Modified non-natural NKG2D ligands that selectively deliver bound heterologous molecules to non-natural NKG2D receptors on CAR cells

Abstract

This application generally relates to the generation of a modified, non-naturally occurring α1-α2 domain of an NKG2D ligand to which a polypeptide with specific target binding properties, e.g., an antibody or variable fragment of an antibody, is bound, for selective delivery to a chimeric antigen receptor (CAR) comprised of a modified, non-naturally occurring NKG2D receptor on a genetically engineered mammalian cell. Targeting of the surface-expressed molecule includes that of a virus-infected cell, which is then attacked and eliminated by genetically engineered cells of the immune system expressing a CAR cognate to the modified, non-naturally occurring α1-α2 domain of the NKG2D ligand to which the polypeptide is bound.

Description

[Technical field]

[0001] The present application generally relates to modified non-natural α1-α2 domains of NKG2D ligands coupled with polypeptides having specific target binding properties, such as antibodies or antibody fragments, that are selectively delivered to chimeric antigen receptors (CARs) comprised of modified non-natural NKG2D receptors on genetically engineered mammalian cells. [Background technology]

[0002] Antibodies (Abs), also known as immunoglobulins (Igs) in many mammals, including humans (Figure 1), are large, Y-shaped proteins used by the immune system to identify and neutralize foreign substances such as bacteria and viruses (Charles Janeway (2001). Immunobiology. (5th ed.), Chapter 3. Garland Publishing. ISBN 0-8153-3642-X. (NCBI (Electronic full text via Bookshelf). Antibodies recognize unique parts of foreign targets, called antigens. Each end of the two arms of the antibody's "Y" contains an antigen-binding site, or paratope (a structure similar to a lock), which is specific for one particular epitope of the antigen (also similar to a key), so that these two structures can bind precisely to each other. Using this binding mechanism, antibodies can tag a microbe or infected cell for attack by other parts of the immune system, or they can directly neutralize their target, for example, by blocking parts of the microbe that are essential for the microbe's invasion and survival. The production of antibodies is the main function of the humoral or "adaptive" immune system. Antibodies are secreted by plasma cells. Antibodies in nature can exist in two physical forms: a soluble form that is secreted from the cell, and a membrane-bound form that is attached to the surface of B cells via the "stem" of the Y.

[0003] Antibodies are glycoproteins that belong to the immunoglobulin superfamily and are typically composed of a basic structural unit with two large heavy chains and two small light chains each. There are several different types of antibody heavy chains and several different types of antibodies that are classified into different isotypes based on which heavy chains they possess. In mammals, five different antibody isotypes are known (Market E, Papavasiliou FN (October 2003) "V(D)J recombination and the evolution of the adaptive immune system". PLoS Biol. 1(1): E16. doi: 10.1371 / journal.pbio.0000016. PMC212695. PMID14551913). Although the general structure of all antibodies is very similar, small regions at the tip of each arm of the Y-shaped protein are highly variable, allowing the existence of millions of antibodies with slightly different tip structures, i.e., antigen binding sites. This region is known as the hypervariable or variable region. Each of these natural variants can bind to a different antigen. This high level of antibody diversity allows the immune system to adapt to and recognize a similarly diverse range of antigens (Hozumi N, Tonegawa S (1976) "Evidence for somatic rearrangement of immunoglobulin genes coding for variable and constant regions". Proc. Natl. Acad. Sci. USA 73 (10): 3628-3632. doi: 10.1073 / pnas. 73.10.3628. PMC431171. PMID 824647.)

[0004] A natural "Y" shaped Ig molecule consists of four polypeptide chains; two identical heavy chains and two identical light chains linked by disulfide bonds. Each heavy chain has two main regions, a constant region (CH) and a variable region (VH). The constant region is essentially identical in all antibodies of the same isotype, but differs in antibodies of different isotypes. The light chain also has two consecutive domains: a smaller constant region (CL) and a variable region (VL) (Woof J, Burton D (2004) "Human antibody-Fc receptor interactions illuminated by crystal structures" Nat Rev Immunol 4 (2): 89-99. doi: 10.1038 / nri1266. PMID15040582).

[0005] Some parts of an antibody have the same function. For example, each of the two arms of the Y has a site that binds to an antigen and thus can recognize a specific foreign substance. This region of the antibody is called the Fv (fragment variable) region. It is composed of one variable region (V) from the heavy chain of the antibody. H ) and one variable region from the light chain (V L (Hochman J, Inbar D, Givol D (1973). An active antibody fragment (Fv) composed of the variable portions of heavy and light chains. Biochemistry 12(6):1130-1135. doi:10.1021 / bi00730a018. PMID4569769). The paratope forms one end of the Fv and is the region that binds to the antigen. This is the V L and V HEach of these six loops is composed of three variable loops of β-strands, which are involved in binding to antigens. These six loops are called complementarity determining regions (CDRs) (North B, Lehmann A, Dunbrack RL (2010). “A new clustering of antibody CDR loop conformations”. J Mol Biol 406(2):228-256. doi:10.1016 / j.jmb.2010.10.030.PMC3065967.PMID21035459).

[0006] Useful polypeptides that retain specific antigen-binding functions can be derived from the CDRs of the variable regions of antibodies. These two antibody variable domains, one light chain (VL) and one heavy chain (VL), each of which has three CDRs, H ) can be fused in tandem in any order using a single short linker peptide of 10 to about 25 amino acids to generate a linear single-chain variable fragment (scFv) polypeptide containing one each of the heavy and light chain variable domains (Bird, RE, Hardman, KD, Jacobson, JW, Johnson, S., Kaufman, BM, Lee, SM, Lee, T., Pope, SH, Riordan, GS, and Whitlow, M. (1988) Single-chain antigen-binding proteins, Science 242, 423-426; Huston, JS, Levinson, D, Mudgett-Hunter, M, Tai, MS, Novotny, J, Margolies, MN, Ridge, R., Bruccoleri, RE, Haber, E., Crea, R., and Opperman, H. (1988). Protein engineering of antibody binding sites: Recovery of specific activity in an anti-digoxin single-chain Fv analogue produced in Escherichia coli.PNAS 85:5879-5883).

[0007] The linker is usually glycine-rich for flexibility, plus serine, threonine, or charged amino acids for solubility, and V H N-terminus of V L or vice versa. This protein retains the specificity of the original immunoglobulin, despite the removal of the constant regions and the introduction of a single linker. This format allows a person skilled in the art of recombinant DNA technology to genetically fuse a linear scFv to the N- or C-terminus of a parent protein in order to confer the antigen-binding properties of the scFv to the parent protein. Many other multivalent and tandem scFv region configurations have been proposed or created, but importantly, all have at least two spatially separated termini, as described below (Le Gall, F.; Kipriyanov, SM; Moldenhauer, G; Little, M (1999) "Di-, tri- and tetrameric single chain Fv antibody fragments against human CD19: effect of valency on cell binding". FEBS Letters 453(1):164-168. doi:10.1016 / S0014-5793(99)00713-9. PMID10403395). Summary of the Invention

[0008] The present disclosure relates to modified α1-α2 domains of NKG2D ligands that bind to heterologous polypeptides, in some embodiments, antibodies or antibody fragments. The modified ligands selectively bind to cognate non-native NKG2D receptors, which in turn selectively bind to their cognate modified ligands. Non-native NKG2D receptors can be expressed on the surface of cells of the immune system, creating chimeric receptors on the surface of the effector cells. The heterologous molecules that bind the ligands can also bind to specific molecules on the surface of target cells, thereby delivering immune effector cells to the target cells. Such effector cells include lymphocytes, B cells, plasma cells, monocytes, macrophages, and dendritic cells.

[0009] In some embodiments, the disclosure relates to an engineered non-natural ligand for an engineered non-natural NKG2D receptor, which has a heterologous molecule attached thereto that selectively binds to an HIV protein present on the surface of an HIV infected cell, and which engineered ligand with the heterologous molecule can selectively bind to the engineered non-natural NKG2D receptor of a CAR cell and cause the destruction of the HIV infected cell.

[0010] In a further embodiment, the HIV protein to which the heterologous molecule selectively binds is an HIV envelope protein.

[0011] In yet another embodiment, the epitope of the envelope protein to which the heterologous molecule selectively binds comprises SEQ ID NO:169 or SEQ ID NO:170.

[0012] In some embodiments of the disclosure, the modified non-natural ligand comprises SEQ ID NO:68, 69, 70, 71, or 72.

[0013] In a further embodiment, the modified non-natural NKG2D receptor comprises SEQ ID NO:54 or 154.

[0014] In some embodiments of the present disclosure, HIV proteins are expressed in HIV-infected cells that have been shocked or activated by mechanisms or agents known to induce expression of HIV proteins in latently HIV-infected cells, i.e., latency-reactivating agents.

[0015] In some embodiments of the present disclosure, the CAR cells are bound to multiple engineered non-natural ligands in which different distinct heterologous molecules bind to different epitopes, proteins, or other molecules on the surface of HIV-infected cells.

[0016] In some embodiments of the present disclosure, a modified non-natural NKG2D receptor comprising SEQ ID NO: 54 or 154 is present on a CAR cell, and the modified NKG2D receptor binds to a modified non-natural ligand consisting of SEQ ID NO: 68, 69, 70, 71, or 72, to which a heterologous molecule or atom that does not bind to an HIV protein is attached.

[0017] In further embodiments, the heterologous molecule or atom modulates a function of the CAR cell, which in yet further embodiments includes proliferation, differentiation, elimination, imaging, antagonism of immunosuppression, homing, or cell lysis of non-HIV infected cells. [Brief description of the drawings]

[0018] The patent or application file contains at least one drawing executed in color. Copies of the patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee. [Figure 1] Diagram of a typical mammalian antibody showing its Y-shaped structure and structural elements. [Diagram 2] Typical CAR anatomical morphology (Gill & June, 2015, excerpt). [Diagram 3]Structure-directed mutagenesis of the α1-α2 domain of MICA to enhance affinity for NKG2D. Structure of the α1-α2 domain of MICA (PDB 1HYR) with the NKG2D-binding interface extensively mutagenized at 57 specific amino acid sites colored in dark grey. [Figure 4] Tyrosine residues Y152 and Y199 in the native NKG2D homodimer. [Diagram 5] Protein sequence alignment of the α1-α2 domains from MICA and ULBP1-6. The amino acids highlighted in grey were selected for NNK mutagenesis in ULBP2 (60 amino acids) and ULBP3 (36 amino acids). The residues highlighted in black were identified as key positions for selection and mutations that modulate binding affinity with NKG2D (Tables 6 and 7). [Figure 6] ELISA results of R3 antibody fusions to non-native α1-α2 domains selected for binding to Y152A NKG2D-Fc. (A) R3 HC25 antibody fusions are not selective for Y152A NKG2D. (B) R3 HC25.17 (SEQ ID NO: 73) antibody fusions are selective for Y152A NKG2D over native NKG2D-Fc. (C) R3 HC.U2RW antibody fusions are not selective for Y152A NKG2D over native NKG2D-Fc. (D) R3 HC.U2S3 (SEQ ID NO: 74) antibody fusions are selective for Y152A NKG2D over native NKG2D-Fc. [Figure 7]Evaluation of effector:target (E:T) cell ratios for killing of HIV-infected primary CD4 T cells by CAR-T cells bearing different concentrations of specifically HIV-targeting MicAbodies. One million primary tonsil-derived cells infected with Bal-GFP R5 virus (approximately 10% infected; 1x104 infected cells) were incubated with 1x105 untransduced CD8 (0:1), or 1x104 (1:1) or 2x105 (20:1) CAR-T cells in the presence of different concentrations of four different broadly neutralizing HIV MicAbodies. After 24 hours, cells were stained and evaluated by flow cytometry. Cells were gated on single / live / CD3+ / CD8- cells expressing or not expressing GFP. Results are shown as an average of triplicates. [Figure 8] Specific killing of primary CD4 cells infected with R5 virus by CAR-T in combination with specific HIV MicAbodies. One million primary tonsil-derived cells infected with Bal-GFP R5 virus (approximately 1x104 infected cells) were incubated with 1x105 CAR-T cells in the presence of different concentrations of HIV-specific MicAbodies or B-cell specific CD20-targeting MicAbodies or HER2-targeting MicAbodies (Her2). After 24 hours, cells were stained and analyzed by flow cytometry. Cells were gated on single cell / live / CD3+ / CD8- and either GFP+ or GFP-. Results are shown as an average of 4 experiments. [Figure 9]Specific killing of primary CD4 cells infected with F4 transmitted / founder virus by CAR-T in combination with specific HIV MicAbodies. One million primary tonsil-derived cells infected with F4-GFP(T / F) virus (approximately 1×104 infected cells) were incubated with 1×105 convertible CAR-T cells in the presence of different concentrations of four distinct HIV-specific MicAbodies, a MicAbody targeting CD20 (Ritux), or a MicAbody targeting HER2 (Her2). After 24 hours, cells were stained and subsequently subjected to flow cytometry. Cells were gated on single cells / live / CD3+ / CD8- and either GFP+ or GFP-. [Figure 10] CAR-T and MicAbody killing of reactivated latently infected reservoir cells from non-viremic patients chronically infected with HIV and receiving ART. CD4+ T cells were isolated by no-touch negative selection from PBMCs collected from known HIV-infected patients receiving ART and reactivated with 100 nM phorbol myristate acetate (PMA) + 1 uM ionomycin for 72 h. Cells were then washed twice and incubated with convertibleCAR-T cells or non-transduced CD8 T cells for 48 h in the presence of an equal concentration of 0.1 nM or 1 nM of a mixture of HIV bNAb-based MicAbodies (3BNC60, 3BNC117, PGT121, and 10-1074), denoted as MIX. Cells were then centrifuged and RNA was extracted from the cell pellet. Cell-associated HIV RNA was measured by ddPCR. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Natural killer (NK) cells of the immune system, cells of the monocyte-macrophage lineage, and certain (CD8+αβ and γδ) T cells play a vital role in humans and other mammals as a first-line innate defense against neoplastic and virus-infected cells (Cerwenka, A., and LLLanier. 2001. NK cells, viruses and cancer. Nat. Rev. Immunol. 1:41-49). NK cells and certain T cells present on their surface NKG2D, a prominent homodimeric surface immune receptor involved in target cell recognition and activation of innate defense against pathological cells (Lanier, LL, 1998. NK cell receptors. Ann. Rev. Immunol. 16:359-393; Houchins JP et al. 1991. DNA sequence analysis of NKG2, a family of related cDNA clones encoding type II integral membrane proteins on human NK cells. J. Exp. Med. 173:1017-1020; Bauer, S et al., 1999. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 285:727-730).Human NKG2D molecules possess a C-type lectin-like extracellular domain that binds to its cognate ligands, 84% of which are identical or homologous to the polymorphic analogs of major histocompatibility complex (MHC) class I chain-related glycoproteins (MIC), monomeric MICA and MICB (Weis et al. 1998. The C-type lectin superfamily of the immune system. Immunol. Rev. 163:19-34; Bahram et al. 1994. A second lineage of mammalian MHC class I genes. PNAS 91:6259-6263; Bahram et al. 1996a. Nucleotide sequence of the human MHC class I MICA gene. Immunogenetics 44:80-81; Bahram and Spies TA. 1996. Nucleotide sequence of human MHC class I MICB cDNA. Immunogenetics 43:230-233). Non-pathological expression of MIC proteins is generally restricted to intestinal epithelium, keratinocytes, endothelial cells, and monocytes, but in response to many types of cellular stress, such as proliferation, oxidation, viral infection, and heat shock, aberrant surface expression of these MIC proteins occurs, characterizing the cells as pathological (Groh et al. 1996. Cell stress-regulated human MHC class I gene expressed in GI epithelium. PNAS 93:12445-12450; Groh et al. 1998. Recognition of stress-induced MHC molecules by intestinal γδT cells. Science 279:1737-1740; Zwirner et al. 1999. Differential expression of MICA by endothelial cells, fibroblasts, keratinocytes and monocytes. Human Immunol. 60:323-330).Pathological expression of MIC proteins is also thought to be involved in several autoimmune diseases (Ravetch, JV and Lanier LL. 2000. Immune Inhibitory Receptors. Science 290: 84-89; Burgess, SJ. 2008. Immunol. Res. 40: 18-34). Differential regulation of NKG2D ligands such as polymorphic MICA and MICB is important to provide the immune system with the means to identify and respond to a wide range of emergency cues while protecting healthy cells from unwanted attacks (Stephens HA, (2001) MICA and MICB genes: can the enigma of their polymorphism be resolved? Trends Immunol. 22: 378-85; Spies, T. 2008. Regulation of NKG2D ligands: a purposeful but delicate affair. Nature Immunol. 9: 1013-1015).

[0020] Viral infection is a common inducer of MIC protein expression, identifying virus-infected cells as targets for attack by NK cells or T cells (Groh et al. 1998; Groh et al. 2001. Co-stimulation of CD8+ αβT-cells by NKG2D via engagement by MIC induced on virus-infected cells. Nat. Immunol. 2:255-260; Cerwenka, A., and LLLanier. 2001).Indeed, to escape such attacks on their host cells, cytomegalovirus and other viruses have evolved mechanisms to prevent the expression of MIC proteins on the surface of infected cells to avoid retaliation from the innate immune system (Lodoen, M., K. Ogasawara, JA Hamerman, H. Arase, JP Houchins, ES Mocarski, and LLLanier. 2003. NKG2D-mediated NK cell protection against cytomegalovirus is impaired by gp40 modulation of RAE-1 molecules. J. Exp. Med. 197:1245-1253; Stern-Ginossar et al., (2007) Host immune system gene targeting by viral miRNA. Science 317:376-381; Stern-Ginossar et al., (2008) Human microRNAs regulate stress-induced immune responses mediated by the receptor NKG2D. Nature Immunology 9:1065-73;Slavuljica,IA Busche,M Babic,M Mitrovic,I Gasparovic,D Cekinovic,E Markova Car,EP Pugel,A Cikovic,VJ Lisnic,WJ Britt,U Koszinowski,M Messerle,A Krmpotic and S Jonjic.2010.Recombinant mouse cytomegalovirus expressing a ligand for the NKG2D receptor is attenuated and has improved vaccine properties.J.Clin.Invest.120:4532-4545).

[0021] Despite that stress, many malignant cells, such as those in lung cancer and glioblastoma brain tumors, also escape expression of MIC proteins and may become particularly aggressive as a result of excessive evasion of the innate immune system (Busche, A et al. 2006, NK cell mediated rejection of experimental human lung cancer by genetic over expression of MHC class I chain-related gene A. Human Gene Therapy 17:135-146; Doubrovina, ES, MM Doubrovin, E Vider, RB Sisson, RJ O'Reilly, B Dupont, and YM Vyas, 2003. Evasion from NK Cell Immunity by MHC Class I Chain-Related Molecules Expressing Colon Adenocarcinoma (2003) J. Immunology 6891-99; Friese, M. et al. 2003. MICA / NKG2D-mediated immunogene therapy of experimental gliomas. Cancer Research 63:8996-9006;Fuertes,MB, MV Girart,LL Molinero,CI Domaica,LE Rossi,MM Barrio,J Mordoh,GA Rabinovich and NW Zwirner.(2008)Intracellular Retention of the NKG2D Ligand MHC Class I Chain-Related Gene A in Human Melanomas Confers Immune Privilege and Prevents NK Cell-Mediated Cytotoxicity. J. Immunology, 180:4606 -4614).

[0022] The high-resolution structure of human MICA bound to NKG2D has been solved, demonstrating that the α3 domain of MICA does not directly interact with NKG2D (Li et al. 2001. Complex structure of the activating immunoreceptor NKG2D and its MHC class I-like ligand MICA. Nature Immunol. 2:443-451; Protein Data Bank accession code 1HYR). The α3 domain of MICA, like that of MICB, is connected to the α1-α2 platform domain by a short, flexible linker peptide, which naturally positions itself as a "spacer" between the platform and the surface of MIC-expressing cells. The three-dimensional structures of the α3 domains of human MICA and MICB are nearly identical (root-mean-square distances of less than 1 Å for 94 C-αα) and are functionally interchangeable (Holmes et al. 2001. Structural Studies of Allelic Diversity of the MHC Class I Homolog MICB, a Stress-Inducible Ligand for the Activating Immunoreceptor NKG2D. J Immunol. 169:1395-1400).

[0023] Certain non-natural α1-α2 domains of NKG2D ligands have been described that have been engineered to bind to the native human NKG2D receptor with higher affinity than the native α1-α2 domain (Candice S Engyel, Lindsey J. Willis, Patrick Mann, David Baker, Tanja Kortemme, Roland K. Strong and Benjamin J. McFarland. Mutations Designed to Destabilize the Receptor-Bound Conformation Increase MICA-NKG2D Association Rate and Affinity. Journal of Biological Chemistry Vol. 282, no. 42, pp. 30658-30666, 2007; Samuel H. Henager, Melissa A. Hale, Nicholas J. Maurice, Erin C. Dunnington, Carter J. Swanson, Megan J. Peterson, Joseph J. Ban, David J. Culpepper, Luke D. Davies, Lisa K. Sanders, and Benjamin J. McFarland. Combining different design strategies for rational affinity maturation of the MICA-NKG2D interface. Protein Science 2012 VOL 21:1396-1402.Herein, we describe non-natural α1-α2 domains of NKG2D ligands that have been modified to bind to non-natural NKG2D receptors that are themselves mutated at sites that result in impaired or lost binding to the natural α1-α2 domain of the NKG2D ligand (David J. Culpepper, Michael K. Maddox, Andrew B. Caldwell, and Benjamin J. McFarland. Systematic mutation and thermodynamic analysis of central tyrosine pairs in polyspecific NKG2D receptor interactions. Mol Immunol. 2011 January; 48(4):516-523; U.S. Patent Application Serial No. 14 / 562,534; U.S. Provisional Patent Application Serial No. 62 / 088,456). The present invention generates bispecific molecules that are composed of specifically engineered non-naturally occurring α1-α2 domains and that specifically target heterologous molecules, including but not limited to heterologous peptides or polypeptides, that bind to a chimeric antigen receptor (CAR), where the receptor for the CAR is composed of a non-naturally occurring NKG2D receptor ectodomain that binds to the engineered α1-α2 domain with higher affinity than the native α1-α2 domain.In turn, such CAR-configured engineered cells of the immune system, e.g., B cells, T cells, NK cells, and macrophages, can overcome many of the known problems of current CAR-T and CAR-NK cell therapeutics, including severe systemic toxicity and antigen avoidance, as described below. (Kalos M, Levine, BL, Porter, DL, Katz, S, Grupp, SA, Bagg, A and June, C. T Cells with chimeric antigen receptors have potent antitumor effects and can establish memory in patients with advanced leukemia. Sci Transl Med 2011;3:95ra73; Morgan RA, Yang JC, Kitano M, Dudley ME, Laurencot CM, Rosenberg SA. Case report of a serious adverse event following the administration of T cells transduced with a chimeric antigen receptor recognizing ERBB2. Mol Ther 2010,18:843-851; Gill and June 2015).

[0024] Using gene transfer techniques, T cells, NK cells, and macrophages can be modified to directly and stably express on their surface the binding domains of antibodies that confer novel antigen specificity (Saar Gill & Carl H. June. Going viral: chimeric antigen receptor T-cell therapy for hematological malignancies. Immunological Reviews 2015. Vol. 263: 68-89; Wolfgang Glienke, Ruth Esser, Christoph Priesner, Julia D. Suerth, Axel Schambach, Winfried S. Wels, Manuel Grez, Stephan Kloess, Lubomir Arseniev and Ulrike Koehl. 2015. Advantages and applications of CAR-expressing natural killer cells. Front. Pharmacol. doi: 10.3389 / fphar.2015.00021). CAR-T cells are an application of this approach that combines the antigen recognition domain of a specific antibody with the intracellular domain of the CD3-ζ chain, the main transmitter of signals from the endogenous T cell receptor (TCR), together with costimulatory molecules such as CD27, CD28, ICOS, 4-1BB, or OX40 into a single chimeric protein, Figure 2. CARs constructed in this way are capable of initiating T cell activation upon binding to a targeted antigen, similar to the endogenous T cell receptor, but in a major histocompatibility complex (MHC)-independent manner.

[0025] As used herein, "soluble MIC protein", "soluble MICA" and "soluble MICB" refer to MIC proteins that contain the α1-α2 domain with or without the α3 domain of the MIC protein, but do not contain membrane-binding motifs, transmembrane domains, or intracellular domains. NKG2D ligands ULBP1-6 do not naturally possess the α3 domain (Cerwenka A, Lanier LL. 2004. NKG2D ligands: unconventional MHC class I-like molecules exploited by viruses and cancer. Tissue Antigens 61(5):335-43. doi:10.1034 / j.1399-0039.2003.00070.x. PMID12753652). The "α1-α2 domain" of an NKG2D ligand refers to the protein domain of the ligand that binds to the NKG2D receptor.

[0026] In some embodiments, the α1-α2 domain of a non-natural NKG2D ligand protein of the invention is at least 80% identical or homologous to the native or naturally occurring α1-α2 domain of an NKG2D ligand, SEQ ID NOs: 1-19. In other embodiments, the modified α1-α2 domain is 85% identical to the native or naturally occurring α1-α2 domain of an NKG2D ligand. In yet other embodiments, the modified α1-α2 domain is 90% identical to the native or naturally occurring α1-α2 domain of a natural NKG2D ligand protein and binds to a non-natural NKG2D.

[0027] The α1-α2 platform domain of a soluble MIC protein is diffusible in the intercellular or intravascular space of a mammal. Preferably, the α1-α2 platform domain of a non-natural MIC protein of the invention is at least 80% identical or homologous to the native or naturally occurring α1-α2 domain of a human MICA or MICB protein and binds to a native NKG2D or, in certain instances, to a modified non-natural NKG2D receptor. In some embodiments, the α1-α2 platform domain is 85% identical to the native or naturally occurring α1-α2 platform domain of a human MICA, human MICB, or human ULBP1-6 protein and binds to a native NKG2D or a modified non-natural NKG2D. In other embodiments, the α1-α2 platform domain is 90%, 95%, 96%, 97%, 98%, or 99% identical to the native or naturally occurring α1-α2 platform domain of a human MICA, human MICB, or human ULBP1-6 protein and binds to a native NKG2D receptor or a modified non-natural NKG2D receptor.

[0028] In some embodiments, a heterologous peptide tag may be fused to the N- or C-terminus of the α1-α2 domain or another soluble MIC protein to aid in purification of the soluble MIC protein. Tag sequences include peptides such as polyhistidine, myc-peptide, or FLAG tags. Such tags can be removed after isolation of the MIC molecule by methods known to those of skill in the art.

[0029] In other embodiments of the invention, specific mutations can be made in the α1-α2 domain of an NKG2D ligand to generate a non-natural α1-α2 domain that binds to a non-natural NKG2D receptor that has itself been engineered to have a lower affinity for the natural NKG2D ligand. This can be done, for example, through genetic engineering. Such modified non-natural NKG2D receptors can be used to generate non-natural NKG2D-based chimeric antigen receptors (CARs) on the surface of NK cells, T cells, macrophages, or stem cells of the immune system that can preferentially bind to, and be activated by, molecules composed of the non-natural α1-α2 domain of the invention. These pairs of non-natural NKG2D receptors and their cognate non-natural NKG2D ligands of the invention will provide important safety, efficacy, and manufacturing advantages for treating cancer and viral infections compared to current CAR-T and CAR-NK cells, as described below.

[0030] The genetic engineering of T cells with CARs has emerged as a promising approach for adoptive T cell therapy against cancer, and CARs targeting many different molecules have been tested in CAR-T cells as therapeutics for malignancies (Porter DL, Levine BL, Kalos M, Bagg A, June CH. Chimeric antigen receptor-modified T cells in chronic lymphoid leukemia. N Engl J Med. 365: 725-733.). Although remarkable clinical efficacy has been observed in hundreds of patients who have undergone adoptive transfer of T cells expressing CD19-specific chimeric antigen receptors, the process of custom genetic engineering of CARs targeting specific antigens, isolation of autologous T cells from patients, genetic engineering of autologous T cells to express personalized CARs, propagation of modified cells in vitro, and quality control of their generation have all been cumbersome and expensive. Currently, this is feasible only in the context of large academic centers with extensive expertise and resources (Gill & June, 2015).

[0031] When autologous CAR-T cells are reinfused into the donor patient, their in vivo proliferation cannot be controlled---a "living therapy" and there is no dose-response relationship with efficacy (Gill & June, 2015). Additionally, tumor escape from CAR T cells can occur through antigen loss escape (Stephan A. Grupp, MD, Ph.D., Michael Kalos, Ph.D., David Barrett, MD, Ph.D., Richard Aplenc, MD, Ph.D., David L. Porter, MD, Susan R. Rheingold, MD, David T. Teachey, MD, Anne Chew, Ph.D., Bernd Hauck, Ph.D., J. Fraser Wright, Ph.D., Michael C. Milone, MD, Ph.D., Bruce L. Levine, Ph.D., and Carl H. June, MD Chimeric Antigen Receptor-Modified T Cells for Acute Lymphoid Leukemia. N Engl J Med 2013;368:1509-1518), this circumvention route could be most easily addressed by sequential treatment with differently targeted CAR-T cells or by initially infusing a T cell product containing CARs with two or more specificities, but would further complicate the manufacturing process and quality control.

[0032] In addition to CAR-T cells that target tumors using single-chain antibody binding domains (scFv), CAR-T cells that use the ligand-binding domain of the NKG2D receptor have been studied in animals and, more recently, in humans (Sentman CL, Meehan KR. NKG2D CARs as cell therapy for cancer. Cancer J. 2014 Mar-Apr;20(2):156-9. doi:10.1097 / PPO.0000000000000029; Manfred Lehner, Gabriel Gotz, Julia Proff, Niels Schaft, Jan Dorrie, Florian Full, Armin Ensser, Yves A. Muller, Adelheid Cerwenka, Hinrich Abken, Ornella Parolini, Peter F. Ambros, Heinrich Kovar, Wolfgang Holter. Redirecting T Cells to Ewing's Sarcoma Family of Tumors by a Chimeric NKG2D Receptor Expressed by Lentiviral Transduction or mRNA Transfection Research Article | published 15 Feb 2012 | PLOS ONE 10.1371 / journal.pone.0031210;www.clinicaltrials.gov NCT02203825).Because expression of NKG2D ligands is increased on the surface of stressed cells, such as tumor cells and virus-infected cells, this family of natural NKG2D ligands has attracted considerable interest as targets for viral infection and cancer immunotherapy (Spear P, Wu MR, Sentman ML, Sentman CL. NKG2D ligands as therapeutic targets. Cancer Immun. 2013 May 1;13:8.; Song DG, Ye Q, Santoro S, Fang C, Best A, Powell DJ Jr., Chimeric NKG2D CAR-expressing T cell-mediated attack of human ovarian cancer is enhanced by histone deacetylase inhibition. Hum Gene Ther. 2013 Mar;24(3):295-305). One NKG2D CAR was a fusion of the full-length NKG2D receptor with CD3ζ (NKG2Dζ); another NKG2D CAR had only the ectodomain of NKG2D fused in the reverse orientation to a second-generation CAR scaffold composed of the transmembrane and intracellular domains from CD28 and the signaling domain of CD3ζ (NKG2D28ζ). Because activation of NKG2D is dependent on the presence of DAP10, CAR-T cells were also constructed in which DAP10 was co-expressed with NKG2Dζ (NKG2Dζ10). T cells expressing any of the above NKG2D CARs responded to stimulation with NKG2D ligands.

[0033] JPEG0007641223000001.jpg1422

[0034] and TNFα and efficiently killed tumor targets expressing NKG2D ligands in vitro (Heather VanSeggelen, Joanne A. Hammill, Anna Dvorkin-Gheva, Daniela GM Tantalo, Jacek M. Kwiecien, Galina F. Denisova, Brian Rabinovich, Yonghong Wan, Jonathan L. Bramson, T cells engineered with chimeric antigen receptors targeting NKG2D ligands display lethal toxicity in mice, Molecular Therapy accepted article preview online 30 June 2015; doi:10.1038 / mt.2015.119). In addition, expression of NKG2D-DAP10-CD3ζ-based CAR significantly enhanced the cytotoxic activity of NK cells against a wide range of tumor subtypes (Yu-Hsiang Chang, John Connolly, Noriko Shimasaki, Kousaku Mimura, Koji Kono, and Dario Campana. Chimeric Receptor with NKG2D Specificity Enhances Natural Killer Cell Activation and Killing of Tumor Cells. Cancer Res; 73(6) March 15, 2013).

[0035] However, after injection into syngeneic mouse hosts, these CAR-T constructs, which bind to and are activated by the natural ligand of the native NKG2D receptor, caused severe toxicity. Signs of toxicity including poor body condition, hunched posture, labored breathing, and reduced core temperature were observed in tumor-bearing and non-tumor-bearing mice treated with NKG2D-based CAR-T cells compared to untreated control mice. The severity of toxicity of NKG2D CAR-T cells varied, with NKG2Dζ10 being severely toxic, NKG2D28ζ showing moderate toxicity, and NKG2Dζ being tolerable. Clinical signs of toxicity and mortality were exacerbated if mice received chemotherapy prior to adoptive transfer of T cells expressing any of the NKG2D CARs (VanSeggelen et al. 2015).Chemotherapy and radiation are known to induce NKG2D ligands in otherwise healthy tissues (Xiulong Xu, Geetha S Rao, Veronika Groh, Thomas Spies, Paolo Gattuso, Howard L Kaufman, Janet Plate and Richard A Prinz. Major histocompatibility complex class I-related chain A / B (MICA / B) expression in tumor tissue and serum of pancreatic cancer: Role of uric acid accumulation in gemcitabine-induced MICA / B expression. BMC Cancer 2011,11:194 doi:10.1186 / 1471-2407-11-194; Gannage M, Buzyn A, Bogiatzi SI, Lambert M, Soumelis V, Dal Cortivo L, Cavazzana-Calvo M, Brousse N, Caillat-Zucman Induction of NKG2D ligands by gamma radiation and Tumor necrosis factor-alpha may participate in the tissue damage during acute graft-versus-host disease. Transplantation. 2008 Mar 27;85(6):911-5. doi:10.1097 / TP.0b013e31816691ef.). Further characterization revealed that this toxicity was accompanied by a systemic cytokine storm and lethal levels of inflammation in the lungs. These data warn that extreme caution must be exercised when using natural NKG2D ligands for targeted immunotherapy, and suggest that enhancing T cell expression of highly activating CARs may be detrimental in vivo (VanSeggelen et al. 2015).

[0036] CAR-T cells, CAR-NK cells, and macrophages composed of non-native NKG2D receptor ectodomains that do not or only weakly bind to native NKG2D ligands do not undergo the above forms of activation and are therefore not as toxigenic as cells expressing CARs based on the native NKG2D receptor. Furthermore, the ectodomain of non-native NKG2D receptors on cells is not downregulated by soluble formats of natural NKG2D ligands or in myeloid-derived suppressor cells (MDSCs) (Deng W, Gowen BG, Zhang L, Wang L, Lau S, Iannello A, Xu J, Rovis TL, Xiong N, Raulet DH, 2015. Antitumor immunity. A shed NKG2D ligand that promotes natural killer cell activation and tumor rejection. Science. 2015 Apr 3;348(6230):136-9. doi:10.1126 / science.1258867. Epub 2015 Mar 5). However, when such a CAR cell bearing the ectodomain of the non-native NKG2D receptor associates with a bispecific molecule bearing the cognate non-native α1-α2 domain of the present invention and its heterologous targeting motif that finds and binds to its intended target, the CAR is activated and the effector function of the CAR cell is expressed. The effector function of CAR-T cells, CAR-NK cells, and CAR-macrophage cells can reduce or impair the viability or function of the targeted cells. The targeted cells can include malignant cells, immunosuppressive cells of tumors, cells that cause autoimmune diseases, such as cells infected with viruses, including but not limited to HIV, hepatitis viruses, HTLV-1, CMV, EBV, and other herpes viruses.

[0037] Since CAR-T or CAR-NK cells composed of non-natural NKG2D receptor ectodomains are not activated except in the presence of associated bispecific molecules composed of the cognate non-natural α1-α2 domains, their activation can be controlled by the administered bispecific molecules, which, as biopharmaceuticals, exhibit pharmacokinetics and pharmacodynamics that are well known in the art. In the event of an adverse event, physicians do not need to deploy an inducible suicide mechanism to destroy the infused CAR cells, as is currently done, but can simply change the dosing regimen of the administered bispecific molecules (Monica Casucci and Attilio Bondanza. Suicide Gene Therapy to Increase the Safety of Chimeric Antigen Receptor-Redirected T Lymphocytes. J Cancer. 2011;2:378-382). Furthermore, such bispecific molecules with different specific targeting motifs can be administered simultaneously or sequentially to help address the occurrence of resistance and evasion of tumor cells or virus-infected cells as a result of loss of the target antigen, without the need to generate, grow, and inject multiple different autologous CAR cells (Gill & June, 2015). The construction of all CARs can be identical for all CAR cells, with the targeting specificity being determined solely by the targeting motif of the bispecific molecule of the invention produced, simplifying the manufacturing process and making it cheaper.

[0038] Many viruses have evolved mechanisms to evade host cell killing by the innate immune surveillance system, especially the NKG2D-dependent components. For example, adenoviruses, cytomegalovirus (CMV), herpesviruses, HIV, human T-cell lymphoma virus-1 (HTLV-1), and papillomaviruses all possess one or more mechanisms. Such viruses can express viral antigens on the surface of infected host cells, whose epitopes can serve as virus-specific molecular targets for binding of antibodies, antibody fragments, or other molecular targeting motifs. These cell surface-exposed molecular targets are attractive targets for antibodies or adoptive cell therapy (ACT) to prevent the spread of viral infection or to treat viral infection by eliminating virus-infected cells.

[0039] Latent HIV-1 infection is established early in acute infection and is found primarily in memory CD4+ T cells. This reservoir is largely nontranscribed but is sufficient to generate infectious virus when host cells are reactivated by antigen or cytokine stimulation or when antiretroviral therapy (ART) is discontinued. Latent HIV reservoirs are found primarily in lymphoid tissues, where >98% of CD4+ T cells reside. ART can suppress viral replication but cannot eradicate the latent reservoir (Ruelas, DS and WC Greene, An integrated overview of HIV-1 latency. Cell, 2013. 155(3): p. 519-29.). Attempts to clear latent HIV-1 initially focused on reactivating the latent provirus using cytokines or T cell receptor activators. However, these strategies resulted in severe side effects and low efficacy. Instead, the so-called “shock and kill” strategy involves reactivating the non-transcribed provirus through the administration of latency-reversing agents (LRAs), compounds capable of inducing HIV-1 transcription (Cary, DC, K. Fujinaga, and BM Peterlin, Molecular mechanisms of HIV latency. J Clin Invest, 2016. 126(2): p. 448-54.). Reactivating the latent virus results in the expression of the HIV envelope glycoprotein gp160 on the surface of activated cells, which is processed into gp120 and gp41. The domains V1, V2, V3, C1, C2, and N segments of gp120 provide attractive targets for attacking HIV-infected cells with neutralizing antibodies and, as described in this invention, CAR-T cells.

[0040] After reactivating latently infected cells, it was predicted that these cells would produce virus (which is stopped by administration of ART) and then die by apoptosis due to the cytopathic effect of the virus, thereby reducing the size of the latent reservoir. Testing this hypothesis showed that reactivated cells do not die, nor does the size of the latent reservoir decrease (Shan L, Deng K, Shroff NS, Durand CM, Rabi SA, Yang HC, Zhang H, Margolick JB, Blankson JN, Siliciano RF, Stimulation of HIV-1-specific cytolytic T lymphocytes facilitates elimination of latent viral reservoir after virus reactivation. Immunity. 2012; 36 (3) p. 491-501.). Two main problems remain after reactivation of the reservoir. The first concerns the emergence of viruses resistant to CTL killing (Deng K, Pertea M, Rongvaux A, Wang L, Durand CM, Ghiaur G, Lai J, McHugh HL, Hao H, Zhang H,, JB, Gurer C, Murphy AJ, Valenzuela DM, Yancopoulos GD, Deeks SG, Strowig T, Kumar P, Siliciano JD, Salzberg SL, Flavell RA, Shan L, Siliciano RF. Broad CTL response is required to clear latent HIV-1 due to dominance of escape mutations. Nature. 2015 Jan 15;517(7534)p.381-5.). This is a common problem in chronically infected individuals (the vast majority of chronically infected individuals) who did not receive ART treatment during the first 6 months of infection.The second problem arises from CTL exhaustion due to exposure to HIV-associated chronic inflammation (Cella M, Presti R, Vermi W, Lavender K, Turnbull E, Ochsenbauer-Jambor C, Kappes JC, Ferrari G, Kessels L, Williams I; CHAVI Clinical Core B, McMichael AJ, Haynes BF, Borrow P, Colonna M; NIAID Center for HIV / AIDS Vaccine Immunology. Loss of DNAM-1 contributes to CD8+ T-cell exhaustion in chronic HIV-1 infection. Eur J Immunol. 2010 Apr; 40(4): p. 949-54.). New approaches to kill reactivated reservoir cells that avoid the problems of viral resistance and cell exhaustion are likely needed. We propose to construct convertible CAR-T cells that utilize broadly neutralizing HIV antibodies to target CTLs to kill reactivated reservoir cells.

[0041] Thus, the present invention overcomes many of these currently recognized challenges of ACT while expanding the versatility and utility of this remarkable and highly promising immunological approach to managing cancer or viral infections using CAR-T cells, CAR-NK cells, and / or macrophages.

[0042] As used herein, "peptide", "polypeptide", and "protein" are used interchangeably; "heterologous molecule", "heterologous peptide", "heterologous sequence", or "heterologous atom" are molecules, peptides, nucleic acid or amino acid sequences, or atoms, respectively, that are not found in nature or normally in physical association with the molecule of interest. As used herein, "non-natural" and "modified" are used interchangeably. As used herein, "natural" and "native" are used interchangeably, and "NKG2D" and "NKG2D receptor" are used interchangeably. The term "antibody" is used herein in the broadest sense and specifically encompasses monoclonal antibodies, multispecific antibodies (e.g., bispecific antibodies), and antibody fragments, so long as they exhibit the desired biological activity. An "antibody fragment" includes a portion of an intact antibody, preferably including its antigen-binding region. Examples of antibody fragments include Fab, Fab', F(ab')2, and Fv fragments; diabodies; linear antibodies; single-chain antibody molecules; and multispecific antibodies formed from antibody fragment(s).

[0043] The term "comprising," which is used interchangeably with "including," "containing," or "featured," is inclusive or open-ended language and does not exclude additional, unrecited elements or method steps. The phrase "consisting of" excludes any element, step, or ingredient not specified in the claim. The phrase "consisting essentially of" limits the scope of the claim to the specified materials or steps and those that do not materially affect the basic and novel characteristics of the claimed invention. The present disclosure contemplates embodiments of the compositions and methods of the invention that correspond to the scope of each of these phrases. Thus, a composition or method that includes recited elements or steps contemplates certain embodiments in which the composition or method consists essentially of or consists of those elements or steps.

[0044] All references cited herein are incorporated by reference in their entirety, whether or not already expressly incorporated. As used herein, the terms "a," "an," and "any" are intended to include both the singular and the plural.

[0045] The present invention has now been fully described, but those skilled in the art will appreciate that it can be practiced within a wide range of equivalent parameters, concentrations, and conditions without departing from the spirit and scope of the invention and without undue experimentation. Although the present invention has been described with reference to specific embodiments thereof, it will be understood that further modifications are possible. This application is generally intended to cover any variations, uses, or adaptations of the present invention in accordance with the principles of the invention, including departures from the present disclosure as are within known or customary practice in the art to which the invention pertains and as applicable to the essential features described herein. EXAMPLES

[0046] Example 1 (Modified α1-α2 domain of NKG2D ligand.) These are examples of conjugating polypeptides to NKG2DLs that have been modified to significantly enhance binding affinity to the human NKG2D receptor. The α1-α2 domain of the MIC protein is an NKG2DL for the NKG2D receptor. This affinity is sufficient to physiologically activate NK cells and stimulate lysis of cells expressing native full-length MIC protein irreversibly tethered to the two-dimensional plasma membrane surface of the "target cell" (Bauer S, Groh V, Wu J, Steinle A, Phillips JH, Lanier LL, Spies T., Science. 1999 Jul 30;285(5428):727-9.). However, since the engineered soluble MIC proteins of the present invention reversibly bind to specific target antigens on the surface of target cells, the binding affinity of the engineered soluble MIC proteins to NKG2D directly affects the stability of the soluble MIC-dependent complex formed between NK cells and cells expressing the target antigen. In particular, if the affinity between sMICA and NKG2D is increased by substantially slowing the dissociation rate or off-rate of the engineered sMICA from NKG2D, the lower the density of soluble MIC molecules bound to the target cells, the greater the NK cell-based killing effect is predicted. Prior to the present invention, no α1-α2 mutations had been identified that altered the killing activity of soluble MIC proteins or significantly reduced the binding off-rate to enhance the affinity of MIC proteins to NKG2D.Computational design efforts showed that three mutations in the α1-α2 domain of wild-type MICA: N69W, K152E, and K154D (WED-MICA) could combine to moderately affect the binding affinity of NKG2D by affecting the stability of unbound MICA and thereby its rate of association to NKG2D or on-rate of binding (Lengyel CS, Willis LJ, Mann P, Baker D, Kortemme T, Strong RK, McFarland BJ. J Biol Chem. 2007 Oct 19;282(42):30658-66. Epub 2007 Aug 8); subsequently, an extensive computational design study by the same group iteratively scanned 22 amino acid positions in MICA that theoretically contact NKG2D according to the published structural description (Li P, Morris DL, Willcox BE, Steinle A, Spies T, Strong RK.,Nat Immunol.2001 May;2(5):443-451), and further rational, iterative computational design of MICA, when combined with the three previously designed changes, experimentally demonstrated that a total of seven compound mutations qualitatively altered the affinity for NKG2D from weak (Kd approx. 2.5μM) to moderately tight (Kd=51nM) (Henager,Samuel H.,Melissa A. Hale,Nicholas J. Maurice,Erin C. Dunnington,Carter J. Swanson,Megan J. Peterson,Joseph J. Ban,David J. Culpepper,Luke D. Davies,Lisa K. Sanders,and Benjamin J. McFarland,2102,Combining different design strategies for rational affinity maturation of the MICA-NKG2D interface.Protein Science 21:1396-1402).In contrast, in the experimental approach described in the present invention, starting from MICA stabilized by the three WED changes of Lengyel et al. (Lengyel CS, Willis LJ, Mann P, Baker D, Kortemme T, Strong RK, McFarland BJ., J Biol Chem. 2007 Oct 19;282(42):30658-66. Epub 2007 Aug 8), amino acid modifications of MICA that slow down the off-rate between the α1-α2 domain of MICA and NKG2D were experimentally selected.

[0047] This example relates to modifying the NKG2D binding affinity of soluble MIC proteins through the introduction of specific mutations at selected amino acid positions within the α1-α2 domain that affect off-rate binding kinetics, thereby altering the NK cell-mediated killing activity of the non-natural targeting MIC molecules of the invention.

[0048] To engineer soluble non-native α1-α2 domains with altered affinity for NKG2D, 57 residues in the α1-α2 domain were selected for extensive mutagenesis ( FIG. 12 ). A synthetic DNA library encoding the α1-α2 domains and containing NNK mutagenic codons at each of the 57 amino acid positions was synthesized and individually cloned as fusions to the pIII minor coat protein of M13 phage, and phage particles displaying the mutagenized α1-α2 variants were produced in SS320 E. coli cells according to standard methodologies (Andris-Widhopf, J., Steinberger, P., Fuller, R., Rader, C., and Barbas, CF, 3rd. (2011) Generation of human Fab antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences, Cold Spring Harbor protocols 2011). The α1-α2 phage library was screened for increased binding affinity using recombinant biotinylated NKG2D as the target antigen, and repeated rounds of deliberately extended binding, extended washing, and elution of phage clones were repeated to select high affinity variants enriched for slow dissociation or off-rates. A set of specific amino acid mutations occurred frequently at six positions in α1-α2 and were selected as favorable amino acid substitutions with enhanced binding affinity for NKG2D (Figure 3, Table 1).

[0049] [Table 1]

[0050] DNA polynucleotides (SEQ ID NOs: 21-24) encoding the α1-α2 domains of four representative variants 15, 16, 17, and 18 containing different combinations of the specific discovered mutations were synthesized (Table 2).

[0051] [Table 2]

[0052] For the NKG2DLs in the above examples, heterologous molecules such as polypeptides were directly linked to each of these four modified α1-α2 NKG2DLs using a linker peptide. The four His-tagged proteins (SEQ ID NOs: 25-28) consisting of modified NKG2DLs with heterologous molecules linked thereto were expressed in insect cells, purified, and characterized for NKG2D binding affinity and kinetic binding parameters. Competitive binding ELISA was used to determine the relative NKG2D binding affinity of the four modified α1-α2 variants. All wells of a maxisorp ELISA plate were coated with soluble wild-type (WT) NKG2DL, sMICA protein to provide a binding partner for the human NKG2D-Fc reagent. Solutions of the four α1-α2 variants and WT and WED-α1-α2 domains (SEQ ID NO: 20) were titrated into the ELISA wells to competitively inhibit the binding of 2 nM human NKG2D-Fc to WT sMICA coated on the plate. The concentration of human NKG2D-Fc bound to WT NKG2DL on the plate was detected using an anti-Fc-HRP antibody. Figure 13, panel A shows that variants 16, 17, and 18 had IC values ​​of 0.7, 0.6, and 0.5 nM. 50 value, whereas variant 15 exhibited an IC 50 Values ​​are presented, all showing significantly better binding to NKG2D than WT NKG2DL, 27-fold, 32-fold, 38-fold, and 11-fold, respectively, and furthermore substantially better than WED-MICA (Table 3).

[0053] [Table 3]

[0054] Importantly, the relative IC 50This difference also translated into better binding to mouse NKG2D-Fc, demonstrating the ability to improve binding of the soluble engineered α1-α2 domain across human and non-human NKG2D receptors, an important property in preclinical drug development.

[0055] To understand the kinetic basis for the changes in affinity, both the on-rates and off-rates for binding of α1-α2 variant NKG2DL to surface-coated biotinylated human NKG2D were measured using Biolayer Interferometry (Octet) at 100 nM of each modified α1-α2 protein. IC 50 Consistent with the ELISA results, variants 16, 17, and 18 each showed a significant reduction in off-rate (18-fold relative to WT), which is the main cause of the increased affinity (approximately 30-fold relative to WT a1-a2; Table 3). Variant 15 showed a slow off-rate similar to 16, 17, and 18, but a reduced on-rate, resulting in stronger affinity than WT but weaker affinity than variants 16, 17, and 18. The mutation at position 125 clearly altered the on-rate, since the only difference between variant 15 (SEQ ID NO:25) and variant 16 (SEQ ID NO:26) was K125N versus K125L, but the reduced off-rate was due to the H161R mutation. Thus, a selected set of NKG2DL mutations (Table 1) was used to increase α1-α2 affinity for NKG2D through a significant decrease in the off-rate, but certain substitutions also altered the on-rate, resulting in a range of incremental affinity increases that were shown in this invention to have differential activity in NK cell-mediated killing assays as described below.

[0056] The ability of the α1-α2 affinity variants to redirect NK cell-mediated lysis of target cells expressing FGFR3 was demonstrated in vitro in a calcein release assay. The human natural killer (NK) cell line NKL was co-cultured with calcein-loaded P815 target cells ectopically expressing FGFR3 and titrated with soluble modified MIC proteins. The results in FIG. 15 showed that the killing activity of FGFR3-specific soluble MIC variants correlated with the engineered α1-α2 affinity. Specifically, variants 16, 17, and 18 showed approximately 15-fold higher killing potency compared to WT at 0.78 nM. WED-MICA (SEQ ID NO: 20) was only slightly better than WT. Thus, the present invention describes amino acid substitutions in the α1-α2 domain that increased the binding affinity of NKG2D by decreasing the off-rate of soluble MIC protein bound to human NKG2D, resulting in increased killing potential as predicted. WED-MICA, which exhibited somewhat higher affinity for NKG2D than WT MICA by increasing the on-rate rather than decreasing the off-rate, did not show significant improvement in target cell killing. Furthermore, WED-MICA showed substantially lower binding to mouse NKG2D than WT MICA, whereas variants 15, 16, 17, and 18 each showed higher affinity for both human and mouse NKG2D.

[0057] These α1-α2 NKG2DL affinity variants 15, 16, 17, and 18 increased the binding affinity of the conjugated polypeptide to the NKG2D receptor, thereby enhancing NK cell-mediated lysis of target cells.

[0058] Example 2. (Non-natural α1-α2 domain of NKG2D ligand and its cognate non-natural NKG2D receptor) The α1-α2 domain of MICA and other NKG2D ligands bind to the NKG2D receptor at known specific sites (Li et al 2001; Benjamin J. McFarland, Tanja Kortemme, Shuyuarn F. Yu, David Baker, and Roland K. Strong. Symmetry Recognizing Asymmetry: Analysis of the Interactions between the C-Type Lectin-like Immunoreceptor NKG2D and MHC Class I-like Ligands. Structure, Vol. 11, 411-422, April, 2003) and drive activation of immune cells bearing the NKG2D receptor, resulting in killing of target cells displaying MICA or other ligands. Using phage display, we engineered the non-native α1-α2 domain of MICA by extensive mutagenesis at 57 specific sites likely involved in binding to NKG2D (Figure 16). A synthetic DNA library encoding the α1-α2 domains and containing an NNK mutagenic codon at each of the 57 amino acid positions was synthesized and individually cloned as fusions to the pIII minor coat protein of M13 phage, and phage particles displaying the mutagenized α1-α2 variants were produced in SS320 E. coli cells according to standard methodologies (Andris-Widhopf, J., Steinberger, P., Fuller, R., Rader, C., and Barbas, CF, 301). rd,2011. Generation of human Fab antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences, Cold Spring Harbor protocols 2011). The α1-α2 phage library was screened for increased binding affinity using recombinant biotinylated NKG2D as the target antigen, and repeated rounds of deliberately extended binding, extended washing, and elution of phage clones were repeated to select high-affinity variants enriched for slow dissociation or off-rates. A set of specific amino acid mutations at nine positions in the α1-α2 domain were selected as preferred sites for amino acid substitutions with enhanced NKG2D binding affinity. We synthesized DNA polynucleotides encoding the α1-α2 domains of eight representative variants (SEQ ID NOs: 29-36) containing different combinations of specific mutations (Table 4).

[0059] [Table 4]

[0060] DNA polynucleotides encoding the α1-α2 domains of the eight variants were amplified using PCR primers (SEQ ID NO: 37-38). Each was subcloned into a His-tagged α1-α2-α3-Fv fusion expression construct (SEQ ID NO: 39) using Blp1 and Sap1 restriction enzymes to replace the sequence encoding the native (wt) α1-α2 sequence with the mutant α1-α2 sequence. The nine fusion proteins (SEQ ID NO: 40-48) were expressed in 293 cells (Expi293™ Expression System, Life Technologies, Thermo Fisher, Inc.) and affinity purified using Ni-affinity chromatography (HisTrap HP, GE Healthcare Life Sciences).

[0061] To construct the NKG2D receptor protein, DNA (SEQ ID NO: 49) encoding the extracellular domain ("ectodomain") of the wild-type receptor was synthesized and the synthetic DNA was cloned into an N-terminal His-avitag expression vector (SEQ ID NO: 78) using PCR primers (SEQ ID NOs: 50-51) and XbaI and BamHI sites. His-avitag-native NKG2D (SEQ ID NO: 52) was transiently expressed in 293 cells and purified using Ni-affinity chromatography. After purification, the NKG2D protein was site-specifically biotinylated using BirA to attach a biotin group to the avitag sequence (BirA biotin-protein ligase standard reaction kit, Avidity, LLC, Aurora, CO.).

[0062] To characterize and compare the kinetic binding parameters of native and eight variant α1-α2 domains to native NKG2D, binding to surface-coated biotinylated native NKG2D ectodomain was measured using Biolayer Interferometry (Octet) with 100 nM of each α1-α2-α3-Fv fusion protein. The results are presented in Table 5.

[0063] [Table 5]

[0064] As shown in Table 5, selected α1-α2 domain mutations as fusions to the heterologous polypeptide α3-Fv of SEQ ID NOs: 42-48 increased the affinity of the α1-α2 domain for native NKG2D through significantly decreasing the off-rate, which was 20- to over 100-fold slower than the wt (SEQ ID NO: 40) and the previously described MICwed α1-α2 domain variant (SEQ ID NO: 41).

[0065] In this example of the invention, as described below, it was further demonstrated that a non-native α1-α2 domain (DSM25, SEQ ID NO:31, Table 4), which has high affinity and a very slow off-rate from native NKG2D (Table 2; SEQ ID NO:43) as an α1-α2-α3-Fv fusion, exhibits tight binding affinity to a non-native NKG2D receptor containing specific mutations that abolish binding to the native NKG2D ligand. Others have demonstrated that mutations at tyrosine 152 and tyrosine 199 of human NKG2D, which are equivalent to positions 73 and 120 in the ectodomain of NKG2D (SEQ ID NO: 49 and FIG. 4), abolish binding to the natural ligand, MICA (David J. Culpepper, Michael K. Maddox1, Andrew B. Caldwell, and Benjamin J. McFarland. Systematic mutation and thermodynamic analysis of central tyrosine pairs in polyspecific NKG2D receptor interactions. Mol Immunol. 2011 January; 48(4):516-523).

[0066] To construct the non-native NKG2D receptor protein, PCR primers (SEQ ID NO: 50-51) were used to clone the DNA encoding the ectodomain of native NKG2D (SEQ ID NO: 49) and insert it into the N-terminal His-avitag expression vector SEQ ID NO: 52 to generate His-avitag-NKG2D (SEQ ID NO: 53). Site-directed mutagenesis was performed on the DNA construct of native NKG2D ectodomain to introduce the mutations Y152A, Y199A, or Y152A+Y199A to generate three non-native variants of human NKG2D (SEQ ID NO: 54-56, respectively). Native NKG2D and the three non-native NKG2D variants with His-avitag were transiently expressed in 293 cells and purified using Ni-affinity chromatography. After purification, the NKG2D protein was site-specifically biotinylated using BirA to bind a biotin group to the avitag sequence (BirA biotin-protein ligase standard reaction kit, Avidity, LLC, Aurora, Colo.).

[0067] To generate fusions of the α3-Fc heterologous polypeptide with the α1-α2 domains of MICwed (SEQ ID NO:29) and DSM25 (SEQ ID NO:31), DNA polynucleotides encoding the α1-α2 domains were amplified using PCR primers (SEQ ID NOs:37-38), and subcloned into the α1-α2-α3-Fc fusion expression construct (SEQ ID NO:57) using XbaI and NcoI restriction enzymes, respectively, to replace the sequence encoding the native (wt) α1-α2 sequence with the mutant α1-α2 sequence. Three fusion proteins, MICA-Fc (SEQ ID NO: 58), MICwed-Fc (SEQ ID NO: 59), and MICv25-Fc (SEQ ID NO: 60), were expressed in 293 cells (Expi293™ Expression System, Life Technologies, Thermo Fisher, Inc.) and affinity purified using Protein A affinity chromatography (Cat. No. 20334, Pierce Biotechnology, Rockford, IL).

[0068] In addition to purifying the three Fc fusion proteins mentioned above, the NKG2D ligand-Fc fusion proteins MICB-Fc, ULBP1-Fc, ULBP2-Fc, ULBP3-Fc, and ULBP4-Fc were purchased from R&D Systems, Inc. (Minneapolis, MN). The binding of the different α1-α2 domain-Fc fusions to both native and non-native NKG2D ectodomain proteins was analyzed using a plate-based ELISA method. All native and non-native α1-α2 domain-Fc fusions were coated overnight at 4°C into separate wells of a Maxisorp 96-well plate using a coating concentration of 2 μg / mL in phosphate-buffered saline (PBS). The plates were washed three times with PBS / 0.05% Tween 20 and blocked with 0.5% bovine serum albumin for 2 h at 20-22°C. Biotinylated native and non-native NKG2D receptor proteins were titrated against plate-bound NKG2D ligands for 2 h at 20-22°C, and after washing three times with PBS / 0.05% Tween 20 at 20-22°C, bound NKG2D proteins were detected using a streptavidin-HRP secondary detection step and developed with 1-Step Ultra TMB Elisa. The native form of the ectodomain of NKG2D (SEQ ID NO: 49) was able to bind all α1-α2 domain-Fc fusions tested. The non-native MIC-v25 α1-α2 domain ligand had the highest affinity (EC 50 =14 nM), which was 8-fold higher than MICwed and over 100-fold higher than all natural α1-α2 domain ligands tested. All ligands containing both natural and non-natural α1-α2 domains lost binding to the Y199A (SEQ ID NO: 55; FIG. 18, panel B) and Y152A+Y199A (SEQ ID NO: 56) double mutant NKG2D receptors. However, of all natural and non-natural α1-α2 domain ligands tested, only the non-natural α1-α2 domain (SEQ ID NO: 31) of MICv25-Fc (SEQ ID NO: 60) retained binding to the Y152A mutant NKG2D ectodomain (SEQ ID NO: 54) with an EC50 of 50 nM.

[0069] The binding specificity of native NKG2D shows a preference for high affinity non-natural ligands, but the strong binding to native NKG2D ligands present in certain healthy tissues and many stressed tissues creates a very high risk of toxicity when using current NKG2D CAR approaches (VanSeggelen et al. 2015). The Y152A non-natural NKG2D receptor specifically bound only to proteins composed of high affinity non-natural α1-α2 domains engineered to have significantly reduced off-rates. This prototypical example highlights the ability of the non-natural α1-α2 domains to bind to the non-natural NKG2D receptor, thus providing selective control of the non-natural NKG2D CAR with bispecific proteins containing the inventive non-natural α1-α2 domains of NKG2D ligands.

[0070] Example 3 (modified α1-α2 domains of NKG2D ligands). This embodiment relates to additional α1-α2NKG2DL affinity variants obtained through genetic engineering of the α1-α2 domain of ULBP proteins. ULBP proteins contain α1-α2 domains that are NKG2D ligands capable of binding to the NKG2D receptor (Cerwenka A, Lanier LL (2004). NKG2D ligands: unconventional MHC class I-like molecules exploited by viruses and cancer. Tissue Antigens 61(5):335-43. doi:10.1034 / j.1399-0039.2003.00070.x. PMID12753652). This affinity of NKG2D binding is sufficient to physiologically activate NK cells and stimulate lysis of cells expressing native full-length ULBP proteins naturally and irreversibly tethered to the two-dimensional plasma membrane surface of "target cells" (Cerwenka A, Lanier LL (2004). NKG2D ligands: unconventional MHC class I-like molecules exploited by viruses and cancer. Tissue Antigens 61(5):335-43. doi:10.1034 / j.1399-0039.2003.00070.x. PMID12753652). However, because the engineered soluble α1-α2 domains fused to heterologous polypeptides in certain embodiments of the invention reversibly bind to specific target antigens on the surface of target cells, the binding affinity of the engineered ULBPα1-α2 domains to NKG2D directly impacts the stability of the artificial synapse formed between NK cells and cells expressing the target antigen, as previously shown by engineered soluble MIC proteins (Example 21-2). To diversify the repertoire of engineered non-natural α1-α2 domains as NKG2D ligands, ULBP proteins were used as substrates or starting points for phage display-based engineering of NKG2D binding affinity.Although structural homology has been observed between ULBPs and MICA (Radaev, S., Rostro, B., Brooks, AG., Colonna, M., Sun, PD. (2001) Conformational plasticity revealed by the cocrystal structure of NKG2D and its class I MHC-like ligand ULBP3. Immunity 15, 1039-49.), the sequence homology of the ULBPα1-α2 domain to MICA was <50%. Therefore, we sought to identify codon positions in the ULBPα1-α2 domain that improve NKG2D binding affinity.

[0071] To engineer soluble non-native α1-α2 domains from ULBP proteins, ULBP2 and ULBP3 were chosen for phage display and selection of mutants with high affinity NKG2D binding. Sixty amino acid positions in the α1-α2 domain of ULBP2 (SEQ ID NO:61) and 36 amino acid positions in the α1-α2 domain of ULBP3 (SEQ ID NO:62) were selected for extensive mutagenesis. Furthermore, conservative cysteine ​​to serine mutations were made at C8S in ULBP2 (SEQ ID NO:61) and C103S in ULBP3 (SEQ ID NO:62) to eliminate the unpaired free cysteine, in order to improve the phage panning process in addition to increasing the stability and function of the NKG2D ligands to which the polypeptides are bound. A synthetic DNA library encoding these cysteine-to-serine altered α1-α2 domains containing an NNK mutagenic codon at each of the selected amino acid positions was synthesized and individually cloned as fusions to the pIII minor coat protein of M13 phage, and phage particles displaying the mutagenized α1-α2 ULBP2 or ULBP3 variants were produced in SS320 E. coli cells according to standard methodologies (Andris-Widhopf, J., Steinberger, P., Fuller, R., Rader, C., and Barbas, CF, 3rd. (2011). Generation of human Fab antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences, Cold Spring Harbor protocols 2011). The α1-α2 phage display library was screened for increased binding affinity to NKG2D using human NKG2D-Fc as the target protein, and repeated rounds of deliberately extended binding, extended washing, and elution of phage clones were repeated to select high affinity variants enriched for slow dissociation rates or off-rates.In ULBP2, specific amino acid mutations at positions R80, V151, V152, and A153 in α1-α2 were frequently observed and identified as favorable amino acid substitutions that enhanced NKG2D binding affinity (Figure 19, panel A; and Table 6).

[0072] [Table 6]

[0073] ULBP3 had specific amino acid mutations at different positions with higher frequency compared to ULBP2. Positions R162 and K165 in the α1-α2 domain of ULBP3 contained specific mutations identified as favorable amino acid substitutions with enhanced NKG2D binding affinity (Table 7). These modified non-native α1-α2 domains from ULBP2 and ULBP3 can be used to enhance NKG2D binding in multiple therapeutic formats, either as single proteins or as fusions with heterologous peptides or polypeptides.

[0074] [Table 7]

[0075] Example 4 (Binding and cell lysis by modified α1-α2 domains of ULBP fused to antibody peptides) The following examples relate to the binding of antibody polypeptides to NKG2DLs engineered to significantly enhance binding affinity to human and mouse NKG2D receptors. The α1-α2 domain of each ULBP protein is the natural ligand of the NKG2D receptor, i.e., NKG2DL. Antibodies are highly stable glycoproteins composed of two large heavy chains and two small light chains (Figure 1). The art lacked an IgG antibody format capable of directly activating immune cells using non-native ULBPα1-α2 domains that bind to the NKG2D receptor more tightly than the native ULBP domain. Furthermore, the ULBPα1-α2 domain provides an alternative NKG2DL for constructing antibody fusions that may have different in vivo properties compared to the MICAα1-α2 domain. For example, in vivo anti-drug antibody responses against the MICAα1-α2 domain in antibody fusions are unlikely to react with or interfere with engineered ULBPα1-α2 domains due to the low sequence homology between the ULBP and MICAα1-α2 domains (Figure 5). This example shows that fusions of engineered ULBPα1-α2NKG2D ligands (Tables 6 and 7) with the heavy chain of an IgG molecule enhance NKG2D binding and target cell killing compared to the native ULBPα1-α2NKG2D ligand, further demonstrating the utility of fusing modified α1-α2 domains to heterologous proteins or peptides.

[0076] To generate fusions of the engineered α1-α2 domains to antibodies, DNA sequences encoding the R80W and V151D (SEQ ID NOs: 63 and 64, respectively) variants of the C8S modified α1-α2 domain of ULBP2 (SEQ ID NO: 61) and the R162G (SEQ ID NO: 65) variant of the C103S modified α1-α2 domain of ULBP3 (SEQ ID NO: 62) were synthesized and cloned as C-terminal fusions into heavy chain sequences from a Her2-specific antibody (Carter, P., Presta, L., Gorman, CM., Ridgway, JB., Henner, D., Wong, WL., Rowland, AM., Kotts, C., Carver, ME., Shepard, HM. (1992) Proc Natl Acad Sci 15, 4285-9.). The resulting fusions were cloned into the mammalian expression vector pD2509 and expressed as paired full IgG antibodies with the parent antibody light chain. Transient expression was performed in HEK293 cells using the Expi293 expression system according to the manufacturer's protocol (Life Technologies) and purified using standard Protein A affinity chromatography. Binding ELISAs performed on ULBP2 and ULBP3 α1-α2 antibody heavy chain fusions showed that the modified ULBP2 fusions (HC_R80W and HC_V151D) and UBLP3 fusions (HC_R162G) bound with higher affinity to human NKG2D compared to the respective native α1-α2 domains fused to the same heavy chain.

[0077] To characterize the target cell killing properties of the engineered ULBP-antibody fusions, the human natural killer (NK) cell line NKL was co-cultured with calcein-loaded SKBR3 target cells expressing Her2 and titrated with the engineered antibody fusion proteins. Results showed that the enhanced cytolytic (killing) activity of Her2-specific non-native ULBP2 and non-native ULBP3 α1-α2 antibody fusions reflects the enhanced affinity of their engineered α1-α2 domains for NKG2D. Specifically, ULBP2 variant fusions HC_R80W and HC_V151D, and ULBP3 variant fusion HC_R162G killed SKBR3 cells more effectively than antibody fusions containing either of the native α1-α2 domains. These data demonstrated that engineered α1-α2 variant-antibody fusions are a universal platform for enabling IgG molecules to tightly bind NKG2D and direct antigen-specific cell lysis.

[0078] Example 5 (Construction of an orthogonal non-native α1-α2 domain that selectively binds to Y152A non-native NKG2D) There is a strong need for a means to selectively control CAR-T cell therapy to reduce toxicity and improve efficacy against tumors (Gill and June, supra). Previous attempts have been made to develop CARs using the ectodomain of CD16 that can subsequently associate through the Fc domain of therapeutic monoclonal antibodies, thus allowing antibody-based control of CAR-T targeting (Chang et al., supra). However, CD16-based CAR-T cells can recognize all endogenous antibody molecules in blood and tissues, and therapeutic antibodies used to control these cells will be subject to interference with the endogenous CD16 receptor on NK cells. Both of these characteristics cause problems related to extratumoral toxicity and poor pharmacokinetics, respectively.

[0079] To address these issues, the inventors have performed genetic engineering to obtain non-native NKG2D CAR-T cells that do not bind to all natural NKG2D ligands and can be controlled through the binding of a high affinity non-native α1-α2 domain, as shown in Example 2. It is further necessary that the non-native α1-α2 domain retains high affinity for the non-native NKG2D and avoids binding to the natural NKG2D domain. Thus, an engineered α1-α2 domain that shows strong selectivity for the non-native NKG2D receptor over the natural NKG2D represents an ideal system to selectively control the non-native NKG2D CAR receptor, or any receptor or protein fused to a non-native NKG2D ectodomain with which the non-native α1-α2 domain can selectively associate.

[0080] The present inventors used phage display to engineer orthogonal non-natural α1-α2 domains that selectively bind to the Y152A NKG2D receptor. As a starting point, three non-natural α1-α2 domains with high affinity for native NKG2D were selected as parent domains for further mutagenesis and screening by phage display. Synthetic DNA libraries were generated for the individual α1-α2 domain variants DSM25, ULBP2 R80W, and ULBP3 R162G (SEQ ID NOs: 31, 63, and 65), whereby the codons for the amino acid residues located in close proximity to the Y152 position of the NKG2D receptor in the bound state were replaced with NNK codons. The DSM25 library consisted of NNK positions at residues 71-75 and 155-159, the ULBP2 R80W library had an NNK codon at positions 154-159, and the ULBP3 R162G library had an NNK codon at positions 155-159. Libraries were cloned as fusions to the pIII minor coat protein of M13 phage, and phage particles displaying the mutagenized α1-α2 domain variants were produced in SS320 E. coli cells according to standard methodologies (Andris-Widhopf, J., Steinberger, P., Fuller, R., Rader, C., and Barbas, CF, 3rd. (2011). Generation of human Fab antibody libraries: PCR amplification and assembly of light- and heavy-chain coding sequences, Cold Spring Harbor protocols 2011). The α1-α2 phage display library was screened for high binding affinity to the non-native Y152A NKG2D receptor by selectively capturing phage clones bound to biotinylated Y152A NKG2D-Fc protein in the presence of non-biotinylated native NKG2D-Fc competitor protein. Selective clones were enriched by repeated rounds of competitive selection with increasing concentrations of non-biotinylated native NKG2D-Fc.

[0081] After four rounds of selection, phage clones were sequenced to identify specific mutations within the NNK mutagenic region. Tables 8, 9, and 10 show selected amino acid residues that were found to be predominant for each α1-α2 domain resulting from selective screening of Y152A NKG2D.

[0082] [Table 8]

[0083] [Table 9]

[0084] [Table 10]

[0085] To confirm that the phage clones displayed appropriate selective binding, phage were generated for individual clones: MICA25.17, MICA25.18, ULBP2.S1, ULBP2.S2, ULBP2.S3, ULBP3.S1, and ULBP3.S2 (SEQ ID NOs: 66, 67, 68, 69, 70, 71, and 72, respectively) and titrated against Y152A or native NKG2D in a binding ELISA. Figure 23, panels A-C demonstrate that all seven phage clones displayed greater than 10-fold selective binding to non-native Y152A NKG2D compared to native or wild-type NKG2D.

[0086] To confirm that the Y152A-selective α1-α2 domain variants retained their specific binding properties in the context of antibody fusions, MICA25.17 and ULBP2.S3 were cloned as C-terminal fusions to the heavy chain of a previously described FGFR3-specific antibody (Qing et al., 2009; SEQ ID NOs: 73 and 74, respectively). The resulting fusions were cloned into the mammalian expression vector pD2509 and expressed as paired full IgG antibodies together with the parent antibody light chain (R3 HC25.17 and R3 HC.U2S3). Transient expression in HEK293 cells using the Expi293 expression system according to the manufacturer's protocol (Life Technologies) and purification was performed using standard Protein-A affinity chromatography. ELISAs measuring the binding of R3 HC25.17 and R3 HC.U2S3 α1-α2 antibody heavy chain fusions to non-native Y152A NKG2D and native NKG2D demonstrated significantly higher binding affinity to Y152A NKG2D compared to native NKG2D (Figure 6, panels B and D). In contrast, antibody fusions to DSM25 and ULBP2 R80W showed preferential binding to native NKG2D-Fc (Figure 6, panels A and C). Taken together, these data demonstrated the invention of a non-native orthogonal α1-α2 domain that retains high affinity binding to the non-native NKG2D receptor and has significantly reduced binding affinity to the native NKG2D receptor. Furthermore, fusions of the orthogonal α1-α2 domain to antibody polypeptides retain their selective binding properties and can be used to redirect non-native NKG2D receptors to new antigens, for example, in the context of CAR-T cells.

[0087] Example 6 (Targeting and killing activity of CAR-T cells bearing a non-native NKG2D ectodomain is controlled using orthogonal α1-α2 domains fused to a targeting antibody). To demonstrate selective control of CAR-T cells constructed with chimeric receptors deploying non-native NKG2D ectodomains, we constructed CARs with either native NKG2D or non-native Y152A NKG2D ectodomains based on previous studies using the 4-1BB / CD3zeta CAR construct (Campana Patent No. 8,399,645) in which the respective NKG2D ectodomain was fused to the CD8 hinge region of the CAR (Figure 2). These constructs were cloned into lentiviral vectors and expressed in primary human CD8+ T cells using lentiviral transduction. The resulting native NKG2D CAR-T cells exhibited specific cell killing activity in vitro, consistent with recognizing the native MICA ligand expressed on target cells. Specifically, the results showed that native NKG2D CAR-T cells killed P1 cells expressing the native MICA ligand, whereas non-native Y152A NKG2D CAR-T cells exhibited a significant loss of ability and a greatly reduced killing of P1 cells expressing MICA. Furthermore, the orthogonal α1-α2 antibody heavy chain fusions R3 HC25.17 and R3 HC.U2S3 selectively activated non-native Y152A CAR-T cells to kill P1 target cells expressing FGFR3, but failed to redirect the killing activity of native NKG2D CAR-T cells. This was in contrast to the R3 HC25 and R3 HC.U2R80W α1-α2 antibody heavy chain fusions, which were not selective for non-native Y152A NKG2D and activated native and non-native CAR-T cells to kill P1 target cells. These data demonstrated that a non-native orthogonal α1-α2 domain engineered to selectively bind to the non-native Y152A NKG2D specifically activated non-native Y152A NKG2D CAR-T cells while avoiding the native NKG2D receptor.

[0088] Example 7 (Construction of an orthogonal non-native α1-α2 domain that selectively binds to Y152A / Y199F non-native NKG2D) It has been demonstrated by others that mutations at tyrosine 152 or tyrosine 199 of human NKG2D, equivalent to positions 73 and 120 of the ectodomain of NKG2D (SEQ ID NO: 49), can greatly reduce binding to the natural ligand, MICA (David J. Culpepper, Michael K. Maddox, Andrew B. Caldwell, and Benjamin J. McFarland. Systematic mutation and thermodynamic analysis of central tyrosine pairs in polyspecific NKG2D receptor interactions. Mol Immunol. 2011 January; 48(4): 516-523). The inventors hypothesized that while mutations at either tyrosine residue had a significant effect on the ability of NKG2D to bind its natural ligands, simultaneous mutations at both tyrosine 152 (Y152) and tyrosine 199 (Y199) would essentially abolish the ability of the receptor to associate with all native ligands. Therefore, we explored individual and combined substitutions of Y152 and Y199 with the aim of identifying both single and double mutant variants that would be unable to associate with any natural ligand, and characterized them for their biochemical behavior. Variants that were successfully expressed and assembled were of particular interest, as these represented inactive ligands that could be more easily generated for analysis.

[0089] The ectodomain of native NKG2D (wild type) (NKG2D.wt, SEQ ID NO: 49) and the ectodomains of candidate non-natural NKG2D variants (SEQ ID NOs: 75-92) (also referred to as "engineered NKG2D" or "eNKG2D") were cloned as fusions into the C-terminus of human IgG1 Fc (without the Fab domain) via a short Factor Xa-recognizable Ile-Glu-Gly-Arg linker (SEQ ID NO: 93) and are interchangeably referred to as Fc-NKG2D.wt or NKG2D.wt and Fc-eNKG2D or eNKG2D (SEQ ID NOs: 94-112). gBlocks® DNA fragments (Integrated DNA Technologies, San Diego, CA) corresponding to the MHCI signal sequence (SEQ ID NO: 113 and 114), human IgG1 Fc with linker (SEQ ID NO: 115), and NKG2D ectodomain variants (SEQ ID NO: 116-124) were synthesized and inserted into pD2610-V12 (ATUM, Newark, CA). DNA constructs exploring substitutions at Y152, Y199, or combinations of Y152 / Y199 mutations (Table 1) were transiently expressed in Expi293™ cells (ThermoFisher Scientific, Waltham, MA) and secreted proteins were purified by Protein A affinity chromatography (catalog no. 20334, Pierce Biotechnology, Rockford, IL). The eluted material was characterized by size exclusion chromatography (SEC) on an Akta Pur Superdex column, and correctly assembled, sized material was fractionated and isolated from the aggregate peak prior to inclusion in the assay.

[0090] Characterization of the purified NKG2D.Y199A-Fc fusion by SEC revealed a composition of predominantly aggregated material (Figure 2). In comparison, both native Fc-NKG2D and Fc-NKG2D.Y152A fusion materials were distinguished by a distinct non-aggregated peak that was easily distinguished from the more rapidly migrating aggregates. The effect of the Y199A mutation on aggregation was also evident in the Y152A / Y199A double mutant Fc-NKG2D fusion variant, indicating that it had the most significant effect on protein misfolding (Figure 2). Thus, this aspect of including Y199A together with any combination of Y152 mutations in the NKG2D variants presented challenges in generating the material required for subsequent genetic engineering attempts, and also raised concerns about assembly and presentation at the cell surface. As a result, we sought to explore other substitutions at Y152 and Y199 that could be combined to obtain more robust molecules. Candidate eNKG2D combination Y152 and Y199 mutants were examined as Fc fusions and are detailed in (Table 1). Furthermore, all purified and expressed Fc-eNKG2D fusion candidates were profiled by SEC, and their chromatograms revealed various levels of aggregate formation (Figures 2 and 3, Table 1). None of the single amino acid substitutions explored at residue 152, alanine, serine, threonine, and valine, affected the assembly of the Fc-NKG2D molecule, whereas Y152-leucine (Y152L) resulted in highly aggregated material. Similar to alanine, neither glutamic acid nor aspartic acid was tolerated at position 199, whereas phenylalanine only slightly increased aggregate formation. Among the mutation combinations explored, Y152A / Y199F, Y152S / Y199F, Y152T / Y199F, and Y152F / Y199F did not adversely affect the desired dimer formation, whereas other combinations increased aggregation.

[0091] Example 8: (Generation of antibody-based bispecific molecules "MicAbody" with non-natural NKG2D ligand variants). To generate non-natural MicA variants fused to human IgG1, for example, DNA polynucleotides encoding the α1-α2 domains of MICwed (SEQ ID NO: 7) and MIC25 (SEQ ID NO: 31) were PCR amplified using primers that also introduced a polynucleotide encoding either the APTSSSGGGGS linker (SEQ ID NO: 135) for fusion to the C-terminal kappa light chain or the GGGS linker (SEQ ID NO: 136) for fusion to the C-terminal heavy chain of human IgG1. In addition, two mutations D265A / N297A (Kabat numbering) that reduce binding to all FcγR receptors were introduced into the CH2 domain of the heavy chain to eliminate antibody-dependent cellular cytotoxicity (ADCC) function (Shields et al., 2001). JBC, 276:6591-6604]. A polynucleotide encoding the α1-α2 domain (SEQ ID NO: 61) of wild-type ULBP2 without GPI linkage (ULBP2.wt) was also cloned and fused to a DNA polynucleotide encoding a linker and to the heavy or light chain of IgG1. These bispecific antibodies are referred to as "MicAbody™" in the singular and "MicAbodies" in the plural, and are bivalent for the fused α1-α2 domain. Examples of antibodies used to generate MicAbodies for the purpose of exploring the genetic engineering of eNKG2D include, but are not limited to, trastuzumab (SEQ ID NO: 137 and 138) and rituximab (SEQ ID NO: 139 and 140), which were subsequently named "trastuzumab-MicAbody" and "rituximab-MicAbody", respectively. Fusion constructs were cloned using Gibson cloning (New England Biolabs, The heavy chains were individually inserted into pD2610-V12 (ATUM, Newark, Calif.) via a plasmid encoding the heavy chain, which was then inserted into pD2610-V12 (ATUM, Newark, Calif.) via a plasmid encoding the light chain fused to a natural or non-natural NKG2D ligand. For a given antibody recognizing a specific antigen, a plasmid encoding the heavy chain and a plasmid encoding the light chain fused to a natural or non-natural NKG2D ligand were co-transfected for transient expression in Expi293™ cells (ThermoFisher Scientific, Waltham, Mass.).Alternatively, plasmids encoding heavy chains fused to natural or non-natural NKG2D ligands were co-transfected with the light chain plasmid. Secreted bispecific antibodies were purified by Protein A affinity chromatography (cat. no. 20334, Pierce Biotechnology, Rockford, IL) and eluted material was characterized by size-exclusion chromatography (SEC) on an Akta Pur Superdex column, fractionation was performed as required. Additionally, SDS-PAGE analysis was performed on purified samples to confirm the expected molecular weights of the fused heavy and fused light chain species.

[0092] Example 9: (Identification of modified NKG2D variants that are unable to bind either natural NKG2D-binding ligands or non-natural ligands with enhanced binding to wild-type NKG2D) The binding affinity of α1-α2 variants to the extracellular domains of native (wild-type) NKG2D and non-native eNKG2D proteins was analyzed using a plate-based ELISA method. SEC-fractionated native Fc-NKG2D and non-native Fc-eNKG2D fusions were each coated overnight at 4°C onto separate wells of a Nunc Maxisorp 96-well plate (Thermo Fisher Scientific, Waltham, MA) using a coating concentration of 1 μg / mL in phosphate-buffered saline (PBS). Plates were washed three times with PBS / 0.05% Tween-20 (PBS-T) at 20-22°C and blocked with 0.5% bovine serum albumin in PBS (PBS-B) for 2 h at 20-22°C. MicAbodies were titrated against plate-bound native or non-native Fc-NKG2D fusions in PBS / 0.5% bovine serum albumin (BSA) / 0.05% Tween-20 (PBS-BT) for 60 min at 20-22°C, washed three times with PBS-T at 20-22°C, and bound bispecific protein was detected using HRP-conjugated anti-human kappa (Abcam, Cambridge MA) in PBS-BT and developed with 1-Step™ Ultra TMB ELISA substrate solution (Thermo Fisher Scientific, Waltham, MA). Binding of ULBP2.wt rituximab-MicAbody (SEQ ID NOs: 139 and 141) discriminated between wild-type NKG2D and eNKG2D variants with reduced binding to the latter, with ligand variants MICwed (SEQ ID NOs: 20 and 78) and MIC25 (SEQ ID NOs: 138 and 80) being more stringent in identifying eNKG2D variants with abolished ligand binding. The binding behavior of each eNKG2D variant to all three dual specific ligands revealed combinations of NKG2D modifications that most greatly reduced binding of wild-type and variant ligands, leading to the identification of a number of novel NKG2D variants with reduced binding to the wild-type and variant ligands. inert It is now possible to select NKG2D variants.

[0093] Additional biophysical analysis of the binding of eNKG2D variants to ligands was also performed using biolayer interferometry (BLI) using a ForteBio Octet system (all from ForteBio LLC, Fremont, CA). For these experiments, human NKG2D ligands MICA-Fc, MICB-Fc, ULBP1-Fc, ULBP2-Fc, ULBP3-Fc, and ULBP4-Fc were purchased from R&D Systems, Inc. (Minneapolis, MN). Ligands in MicAbody format were captured on anti-human IgG Fc capture (AHC) biosensor chips. After a baseline was established, the chips were exposed to a titration series of Fc-eNKG2D fusion proteins ranging from 300 nM to 0.41 nM, and association / dissociation kinetics were monitored. All steps were performed in PBS-BT. Fc-eNKG2D fusion protein was then captured on an AHC chip and MicAbody titrated to characterize binding kinetics.

[0094] To determine the maximum response defined by the binding of native NKG2D to either MICwed or MIC25, native Fc-NKG2D fusions were captured on AHC biosensors and incubated with 20 nM trastuzumab-MICwed MicAbody or 20 nM trastuzumab-MIC25 MicAbody for 2 min, followed by observing dissociation kinetics for 30 s. Binding analysis was then performed under the same conditions using Fc-eNKG2D fusion receptor as capture agent, and the binding level of each eNKG2D was ranked as a percentage of the maximum binding response established by Fc-NKG2D.wt (Table 2). For MICwed, the response of all single mutant Fc-eNKG2D variants was reduced to 50%, except for Y199F, which maintained a 100% binding response. However, all double mutant Fc-eNKG2D variants completely abolished binding to MICwed. For MIC25, all single mutant Fc-eNKG2D variants and Y152V / Y199F maintained 100% binding response compared to wild-type Fc-NKG2D binding, but several of the double mutant Fc-eNKG2D variants, including Y152A / Y199F, Y152S / Y199F, and Y152T / Y199F, showed a decrease in binding by up to 50%.

[0095] ELISA assays using Fc-eNKG2D fusions as capture agents were performed with ULBP2.wt, MICwed, MIC25 MicAbody and titrated starting at 300 nM. When possible, ECs were calculated using GraphPad Prism. 50 The values ​​were calculated (Table 11).

[0096] [Table 11]

[0097] Native NKG2D bound to ULBP2, MICwed, and MIC25-based MicAbodies with affinities calculated to have Kds values ​​of 1.4, 0.007, and 0.005 nM, respectively. The affinity of the ULBP2 and MICwed MicAbodies to all single mutant eNKG2D candidates was reduced, while binding to the MIC25 eNKG2D candidate was retained. However, all double mutant eNKG2D candidates showed abolished or significantly reduced binding to all three ligands - ULBP2, MICwed, and MIC25, in the Micabody format.

[0098] The eNKG2D variants eNKG2D5 (Y152A / Y199F), eNKG2D7 (Y152S / Y199F), eNKG2D8 (Y152T / Y199F), and eNKG2D9 (Y152V / Y199F) had reduced or abolished binding to ULBP2, MICwed, and MIC25-based MicAbodies by both Octet analysis and ELISA (Tables 2 and 3). Furthermore, eNKG2D5, 7, and 8 had the least amount of aggregation, suggesting more robust protein assembly upon 293T expression (Table 1). eNKG2D5 (SEQ ID NO: 102) was examined in more detail for binding to wild-type ligand as a MicAbody captured on an Octet AHC chip. The single mutant Fc-NKG2D.Y152A (SEQ ID NO: 95) had reduced binding to all natural ligands compared to native (SEQ ID NO: 94) NKG2D (Figure 5). The response curve for binding of eNKG2D5(Y152A / Y199F) was even more depressed compared to Y152A eNKG2D. eNKG2D5(Y152A / Y199F, hereafter referred to as "AF" or "NKG2D.AF") was selected as the lead NKG2D variant for engineering alloselective orthogonal non-natural ligands.

[0099] Example 10: (Construction of an orthogonal non-natural α1-α2 domain that selectively binds to a non-natural NKG2D.AF ectodomain). Using phage display, we engineered an orthogonal non-natural α1-α2 domain that selectively binds to the NKG2D.AF (SEQ ID NO: 102) receptor. As a starting point, the non-natural ULBP2.R80W α1-α2 domain (FIG. 1B; SEQ ID NO: 142), which has high affinity for the native wild-type NKG2D (NKG2D.wt) ectodomain, was selected as the parent domain for further mutagenesis and screening by phage display. A synthetic DNA library was generated for the α1-α2 domain of ULBP2.R80W (SEQ ID NO: 108), which additionally carries a C8S mutation to eliminate possible disulfide bonds. Codons for the amino acid residues of the ligand that are located in close proximity to positions Y152 and Y199 of the native NKG2D receptor in the bound state were replaced with NNK codons; the library consisted of NNK codons at positions 154 to 159. Libraries were cloned as fusions to the pIII minor coat protein of M13 phage, and phage particles displaying the mutagenized α1-α2 domain variants were produced in SS320 E. coli cells following standard methodologies (Andris-Widhopf, J., Steinberger, P., Fuller, R., Rader, C., and Barbas, CF, 3rd. (2011). These α1-α2 phage display libraries were screened for high binding affinity to the non-native NKG2D.AF receptor by selectively capturing phage clones bound to biotinylated Fc-NKG2D.AF protein in the presence of non-biotinylated native Fc-NKG2D.wt competitor protein. Selective clones were enriched by multiple rounds of competitive selection with increasing concentrations of non-biotinylated native Fc-NKG2D.

[0100] After four rounds of selection, phage clones were individually arrayed in a 96-well format and spot ELISA was performed to confirm preferential differential binding to non-native NKG2D.AF bound to the plate compared to NKG2D.wt. Bound phage were detected using biotinylated M13 phage coat protein monoclonal antibody E1 (ThermoFisher Scientific, Waltham, MA), streptavidin-HRP detection (R&D Systems, Minneapolis, MN), and 1-Step Ultra TMB ELISA development (ThermoFisher Scientific, Waltham, MA). The spot ELISA signal for each clone was expressed as the ratio of phage binding to NKG2D.AF to phage binding to NKG2D.wt. Phages with a ratio of 14 or greater were sequenced to identify specific mutations within the NNK mutagenized region. When multiple clones representing the same sequence were identified, the ratios of ELISA signals were plotted and the consistency of the phage clones was confirmed by clustering of data points (data not shown).

[0101] Thirty of the variants identified by ELISA were grown in individual monocultures to generate high-titer microbatches of phage. Purified phage concentrations were determined by OD 268 = 0.5 and then subjected to a 1:3 dilution series against plate-bound Fc-NKG2D.AF or Fc-NKG2D.wt, and phage detection and ELISA development were performed as described above. All 30 variants assayed in this manner consistently showed selective binding to NKG2D.AF, with little or no binding to NKG2D.wt, even at the highest concentrations of phage assayed. Selected phage also showed a 2 log or greater shift in phage concentration to achieve half-maximal binding between NKG2D.AF and NKG2D.wt.

[0102] To confirm that the NKG2D.AF selective α1-α2 domain variants retained specific binding properties in the context of an antibody fusion, 21 variants (Table 5; e.g., SEQ ID NOs: 143-150) were cloned as C-terminal fusions with an APTSSSGGGGS linker to the light chain of the rituximab antibody. The resulting fusions were cloned into the mammalian expression vector pD2610-V12 (ATUM, Newark, CA) via Gibson cloning (New England Biolabs Inc., Ipswich, MA) and co-expressed with the parent antibody heavy chain as paired full IgG antibodies. Transient expression was performed in Expi293™ cells (ThermoFisher Scientific, Waltham, MA) according to the manufacturer's protocol and purified using standard Protein A affinity chromatography (catalog no. 20334, Pierce Biotechnology, Rockford, IL). ELISAs measuring binding of each variant ULBP2 α1-α2 antibody fusion to non-native Fc-NKG2D.AF and native Fc-NKG2D.wt demonstrated significantly higher binding affinity to NKG2D.AF compared to native NKG2D.wt (Table 12).

[0103] [Table 12]

[0104] Taken together, these data demonstrate the invention of non-native orthogonal α1-α2 domains that possess high binding affinity to the non-native NKG2D.AF receptor and significantly reduced binding affinity to the native NKG2D receptor. Furthermore, fusions of these orthogonal α1-α2 domains to antibody polypeptides retain their selective binding properties and can be used, for example, in the context of chimeric antigen receptor (CAR) T cells to redirect the non-native NKG2D.AF receptor towards a specific antigen.

[0105] Example 11: (Identification of non-natural NKG2D ligands capable of discriminating between non-natural NKG2D receptor variants by selectively binding to one or the other) Phage display to engineer an orthogonal non-native α1-α2 domain that selectively binds to NKG2D.Y152A (hereafter referred to as the NKG2D.YA, receptor) was performed as described above using the non-native ULBP2.R80W α1-α2 domain (SEQ ID NO: 142) as a starting point. The α1-α2 phage display library was panned for high binding affinity to the non-native Fc-NKG2D.YA receptor by selectively capturing phage clones that bound to biotinylated Fc-NKG2D.YA (SEQ ID NO: 95) protein in the presence of non-biotinylated native Fc-NKG2D.wt (SEQ ID NO: 94) competitor protein. Further phage clone validation studies resulted in the identification of variants that preferentially bound to Fc-NKG2D.YA compared to Fc-NKG2D.wt (Table 13).

[0106] [Table 13]

[0107] For example, ULBP2.S3 (SEQ ID NO: 151) consistently demonstrated selective binding to non-native NKG2D.YA compared to native NKG2D.wt by ELISA and Octet analysis (both monomeric His-tagged and bispecific antibody fusion formats). This represented a distinct form of the invention of a non-native orthogonal α1-α2 domain (in this case NKG2D.YA as opposed to NKG2D.AF of Example 2) with high affinity binding to the non-native NKG2D receptor. Furthermore, fusions of the orthogonal α1-α2 domain to an antibody polypeptide retained its selective binding properties and were used to selectively redirect the non-native NKG2D receptor towards a specific molecule determined by the fused heterologous peptide, such as an antibody.

[0108] To determine whether the non-natural α1-α2 domains that selectively bind NKG2D.YA (ULBP2.S3, SEQ ID NO: 151) and NKG2D.AF can discriminate between these two non-natural receptor variants, a titration ELISA was performed. For all 21 of the selected α1-α2 variants that bind to NKG2D.AF, binding to NKG2D.AF was directly compared to binding to NKG2D.YA. Four of these exhibited the following characteristics: an inability to bind NKG2D.wt, strong affinity for NKG2D.AF, and significantly reduced (15-20 fold) or no binding to NKG2D.YA compared to NKG2D.AF.These four non-natural ULBP2α1-α2 variants, ULBP2.C, ULBP2.R, ULBP2.AA, and ULBP2.AB (SEQ ID NOs: 143, 145, 147, and 149), were analyzed using the NetMHC 4.0 server (to search for peptide-MHC class I binding to all HLA supertype representatives using 9mer peptide analysis; http: / / www.cbs.dtu.dk / services / NetMHC / ) and the NetMHCII 2.3 server (to search for peptide-MHC class II binding to HLA-DR, HLA-DQ, and HLA-DP haplotypes using 15mer peptide analysis; http: / / www.cbs.dtu.dk / services / NetMHCII / ), both of which are algorithms developed by the Technical University of Denmark (http: / / www.bioinformatics.dtu.dk / ; Andreatta M and Nielsen M, Gapped sequence alignment using artificial neural networks: application to the MHC class I system, 2016 Bioinformatics, 32:511, PMID:26515819; Jensen KK, Andreatta M, Marcatili P, Buus S, Greenbaum JA, Yan Z, Sette A, Peters B, and Nielsen M, Improved methods for predicting peptide binding affinity to MHC class I molecules, 2018 Immunology, PMID:29315598) were also used to examine changes in the predicted immunogenicity profile compared to the wild-type ULBP2 peptide sequence (SEQ ID NO:61). Mutations engineered into ULBP2.C, ULBP2.R, and ULBP2.AB did not increase predicted immunogenicity, whereas the immunogenicity of ULPB2.AA was slightly increased for some haplotypes (Figures 8 and 9).As a result of the specificity of ULBP2.R for NKG2D.AF and its predictable lack of immunogenicity, ULBP2.R was selected for further ELISA analysis to directly compare its binding behavior to ULBP2.S3 (a non-natural orthogonal ligand selected for NKG2D.YA), ULBP2.R80W (a non-natural ligand with enhanced affinity for wild-type NKG2D), and wild-type ULBP2 (ULBP2.wt). The binding of four rituximab-MicAbody reagents (SEQ ID NOs: 139 and 151, 139 and 152, 153 and 140, and 139 and 141 as the heavy and light chains for ULBP2.R, ULBP2.S3, ULBP2.R80W, and ULBP2.wt, respectively) was assayed against wild-type NKG2D (NKG2D.wt) and two inactive non-natural variants, NKG2D.YA and NKG2D.AF. The data showed that the variant ULBP2.S3 selected for NKG2D.YA as a MicAbody bound with high affinity to NKG2D.YA but did not associate with NKG2D.AF or native NKG2D. Furthermore, the variant ULBP2.R selected for NKG2D.AF in the MicAbody format bound with high affinity to NKG2D.AF but did not associate with NKG2D.YA or native NKG2D. These results demonstrated tremendous potential for the exploration of the NKG2D-MIC ligand axis and for the development of unique pairs of novel selective non-natural NKG2D receptors and their respective cognate non-natural MIC ligand binding partners.

[0109] Example 12: (Targeting and killing activity of CAR-T cells expressing a non-native NKG2D.AF ectodomain is controlled by an orthogonal α1-α2 domain fused to a heterologous targeting polypeptide) There is a strong need for a means to selectively control CAR-T cell therapy to reduce toxicity and improve efficacy against tumors (Gill and June, supra). Previous attempts have been made to develop CARs using the ectodomain of CD16 that can subsequently be associated through the Fc domain of therapeutic monoclonal antibodies, thus enabling antibody-based control of CAR-T targeting (Chang et al., supra). However, CD16-based CAR-T cells can recognize almost all endogenous antibody molecules in blood and tissues, and therapeutic antibodies used to control these cells will be subject to competition from endogenous CD16 receptors on NK cells, PMNs, monocytes, and macrophages. Both of these characteristics contribute to the problems of extratumoral toxicity and poor pharmacokinetics, respectively.

[0110] Natural NKG2D ligands are present in certain healthy tissues and many stressed tissues, resulting in a very high risk of toxicity when using current NKG2D CAR approaches (VanSeggelen et al. 2015). The Y152A non-natural NKG2D receptor specifically bound to the non-natural α1-α2 domain of the NKG2D ligand, constituting one example of a means by which the activity of non-natural NKG2D CARs can be selectively controlled using bispecific proteins composed of the inventive non-natural α1-α2 domain of the NKG2D ligand.

[0111] The inventors engineered CAR-T cells with a receptor composed of the modified Y152A / Y199F ("AF") ectodomain of NKG2D, which is orthogonal to all native NKG2D ligands or the previously described Y152A modified NKG2D and does not bind to the cognate non-native α1-α2 domain (NKG2D.YA). The cognate non-native α1-α2 domain of the invention binds with high affinity to the non-native NKG2D.AF ectodomain and avoids binding to native NKG2D ectodomain and NKG2D.YA ectodomain. Thus, the engineered α1-α2 domain, which shows strong selectivity for the non-native NKG2D.AF ectodomain over native NKG2D and non-native NKG2D.YA, represents an ideal system for selectively controlling the non-native NKG2D CAR receptor, or any receptor or protein fused to a non-native NKG2D ectodomain with which the non-native α1-α2 domain of the invention can selectively associate. The invention further provides for the expression in a single cell of two distinct CARs, one comprised of NKG2D.YA and the other of NKG2D.AF, each signaling through distinct intracellular domains, with dual independent control of cellular activity by extracellular exposure to their respective cognate orthogonal MicAbodies or separate non-antibody fusion polypeptides.

[0112] To demonstrate selective control of CAR-T cells constructed with chimeric receptors deploying non-native NKG2D.AF ectodomains, CARs (SEQ ID NOs: 155, 157, 159) bearing either native NKG2D.wt (SEQ ID NO: 49), non-native NKG2D.YA (SEQ ID NO: 54), or non-native NKG2D.AF (SEQ ID NO: 154) ectodomains were constructed based on previous studies using the 4-1BB / CD3zeta CAR construct (Campana Patent No. 8,399,645), in which the respective NKG2D ectodomains were fused to the CD8 hinge region of the CAR. These constructs (SEQ ID NOs: 156, 158, 160) were cloned into lentiviral vectors and expressed in primary human CD8+ T cells using lentiviral transduction. HeLa cells have constitutively upregulated concentrations of MIC ligands, including MICA, MICB, ULBP3, and ULBP2 / 5 / 6, on their surface (the antibody used to confirm this cannot distinguish between these three ULBPs; Human ULBP-2 / 5 / 6 Antibody, R&D Systems, Minneapolis, MN). HeLa cells were also transfected to overexpress either native ULBP1 or the NKG2D.AF-selected variant ULBP2.R on their surface, and these cells were used as targets for in vitro killing assays. HeLa target cells were preloaded with calcein and exposed to NKG2D.wt-CAR, NKG2D.YA-CAR, or NKG2D.AF-CAR CD8 cells at increasing effector-to-target ratios (E:T) for 5 hours, after which the amount of calcein released into the supernatant was quantified and normalized to the total calcein released upon detergent treatment. Since high levels of MIC ligands are naturally expressed on the surface of HeLa cells, CD8 cells expressing native NKG2D (NKG2D.wt) as a CAR were able to associate with HeLa cells via this overexpressed native ligand and exert cytolytic activity. However, CD8 cells transduced with both NKG2D.YA-CAR and NKG2D.AF-CAR demonstrated little lysis of native HeLa cells even at high E:T ratios, and showed activity levels comparable to untransduced CD8 T cells.When ULBP1 was overexpressed on the surface of HeLa cells, only NKG2D.wt-CAR CD8 T cells significantly lysed HeLa cells. NKG2D.YA-CAR cells, but not NKG2D.AF-CAR cells, showed some additional killing at high E:T ratios, indicating that the double mutation Y152A / Y199F inactivates NKG2D more than the single Y152A mutation. In HeLa cells overexpressing the NKG2D.AF-selective non-native ULBP2.R, NKG2D.wt-CAR cells directed lysis (by recognition of the endogenous MIC ligand), whereas NKG2D.AF-CAR cells directed significant levels of lysis consistent with the receptor engaging its selective ligand.

[0113] To demonstrate that only the appropriate cognate targeting MicAbodies can direct the lysis of NKG2D.YA-CAR cells or NKG2D.AF-CAR cells, Ramos cells were used as targets for cell lysis in combination with Rituximab-based MicAbodies linked to either non-natural ULBP2.S3 or ULBP2.R orthogonal ligands. The Rituximab-ULBP2.S3 MicAbody was able to direct the cell killing activity of NKG2D.YA-CAR CD8 cells but not NKG2D.AF-CAR cells, whereas the Rituximab-ULBP2.R MicAbody was able to direct the activity of NKG2D.AF-CAR cells but not NKG2D.YA-CAR cells. This further demonstrates the selectivity of the two non-natural ULBP2 variants over their cognate non-natural NKG2D variants engineered as preferred partners. To demonstrate the specificity of the antibody portion of the MicAbody, an in vitro killing assay was performed with NKG2D.AF-CAR CD8 cells pre-armed by incubation with either Rituximab-ULBP2.R, Trastuzumab-ULPB2.R (SEQ ID NOs: 95 and 133, heavy and light chains, respectively), or an equimolar combination of the two, at saturating total concentrations of MicAbody. After removing unbound MicAbody by washing, CD8 cells were applied to Ramos cells (expressing CD20, the target of Rituximab) or CT26-Her2 (a murine cell line transfected to express human Her2) pre-loaded with calcein. After 2 hours of incubation at two different E:T ratios, the amount of calcein released was quantified. When cells were pre-armed with Rituximab-MicAbody, only Ramos cells were lysed, whereas Trastuzumab-MicAbody directed cytolytic activity only against CT26-Her2 cells. However, rituximab- and trastuzumab-ULBP2.R MicAbody was used to treat NKG2D.AF-CAR CD8 cells. bothWhen pre-armed simultaneously, both target cell lines were lysed, demonstrating that these CAR cells can be easily multiplexed by engineered selective and privileged receptor-ligand partners and thereby directed to simultaneously engage distinct tumor targets.

[0114] Example 13: (Killing of productively HIV-infected human tonsil CD4 T cells) CD8+ T cells were isolated from PBMCs of healthy donors, activated with anti-CD3 / CD28 beads, and expressed as CD8 T cells expressing inactive NKG2D, CD8 hinge and transmembrane domains, costimulatory 4-1BB domain, and

[0115] JPEG0007641223000015.jpg1625

[0116] The CAR-T cells were transduced with a CAR composed of NKG2D, a marker for HIV-1 NKG2D activation, which we call ConvertibleCAR cells. These ConvertibleCAR-T cells were only able to indirectly bind to a broadly neutralizing HIV antibody fused to a modified non-natural ligand that was cognate to the inactive NKG2D receptor of ConvertibleCAR. As a negative control, non-transduced CD8 T cells from the same donor were also prepared in parallel. Based on the sequences of the broadly neutralizing antibodies 3BNC60, 3BNC117, PGT121, and 10-1074, four HIV-specific MicAbodies were generated (SEQ ID NOs: 161 and 162 (3BNC60), the heavy and light chains of MicAbody, respectively; 163 and 164 (3BNC117), the heavy and light chains of MicAbody, respectively; 165 and 166 (PGT121), the heavy and light chains of MicAbody, respectively; 167 and 168 (10-1074), the heavy and light chains of MicAbody, respectively). These MicAbodies bind to specific epitopes on the HIV gp160 envelope molecule. The targeted epitope bound by 3BNC60 and 3BNC117 is sequence number 169, and the targeted epitope bound by PGF12 and 10-1074 is sequence number 170 (Deng K, Pertea M, Rongvaux A, Wang L, Durand CM, Ghiaur G, Lai J, McHugh HL, Hao H, Zhang H,, JB, Gurer C, Murphy AJ, Valenzuela DM, Yancopoulos GD, Deeks SG, Strowig T, Kumar P, Siliciano JD, Salzberg SL, Flavell RA, Shan L, Siliciano RF Broad CTL response is required to clear latent HIV-1 due to dominance of escape mutations. Nature. 2015 Jan 15; 517 (7534) p. 381-5.). As negative controls, MicAbodies targeting CD20 or HER2 were also developed.

[0117] Human tonsil cells from four healthy donors were processed to generate human lymphocyte aggregate cultures (HLAC). HLAC cells were overlaid onto 293T cells that had been previously transfected with DNA corresponding to R5-tropic HIV-1 and a GFP reporter gene. After 24 hours, HLAC cells were removed and HIV infection was allowed to continue to spread for another 4 days. GFP-positive infected HLAC cells were then exposed to untransduced CD8 T cells or ConvertibleCAR-T cells armed with the indicated MicAbody and cultured for 48 hours in the presence of 5 μM saquinavir to prevent further viral spread. Cells were then harvested by centrifugation, washed, stained, and viability was assessed in infected and non-infected cells using an LSRII flow cytometer.

[0118] Evaluation of effector:target (E:T) cell ratios for killing of HIV-infected primary CD4 T cells by CAR-T cells with different concentrations of a specifically HIV-targeting MicAbody. As above, one million primary tonsil-derived cells (approximately 10% infected, 1 × 10 4 infected cells) in the presence of different concentrations of four different broadly neutralizing HIV MicAboves. 5 untransduced CD8 (0:1), or 1 × 10 4 Pieces (1:1) or 2×10 5The CAR-T cells were incubated with 1000 cells / mL (20:1) of HIV-specific MicAbody and 1000 cells / mL (1:1) of CAR-T cells. After 24 hours, the cells were stained and evaluated by flow cytometry. The cells were gated on single / live / CD3+ / CD8- cells expressing or not expressing GFP. The results, averaged over three experiments, are shown in Figure 7. In these studies, the combination of HIV-specific MicAbody and ConvertibleCAR-T cells specifically killed tonsil cells infected with R5 HIV virus. The optimal effector:target ratio for killing ranged from 1:1 to 10:1, and there was no reduction in the viability of uninfected cells. Killing was highly restricted to infected cells, i.e., cells expressing GFP. GFP- cells present in the same cultures showed little or no reduction in cell number (Figures B and C; GFP+ vs. GFP-). Furthermore, when non-transduced CD8 T cells from the same donor or MicAbodies not targeting HIV (e.g., MicAbodies targeting CD20 or MicAbodies targeting Her2) were used, no killing of uninfected cells occurred, nor did killing of infected cells.

[0119] Specific killing of primary CD4 cells infected with R5 virus by CAR-T in combination with a specific HIV MicAbody. One million primary tonsil-derived cells (approximately 1 × 10 4 infected cells) in the presence of different concentrations of HIV-specific MicAbodies or B cell-specific CD20-targeting MicAbodies or HER2-targeting MicAbodies (Her2), at 1 × 10 5 The cells were incubated with 10 CAR-T cells. After 24 hours, the cells were stained and analyzed by flow cytometry. Cells were gated on single cells / live / CD3+ / CD8- and either GFP+ or GFP-. The results, averaged over four experiments, are shown in Figure 8.

[0120] Specific killing of primary CD4 cells infected with F4 transmitted / founder virus by CAR-T in combination with a specific HIV MicAbody. One million primary tonsil-derived cells (approximately 1 × 10 4 infected cells) in the presence of different concentrations of four separate HIV-specific MicAbodies, a MicAbody targeting CD20 (Ritux), or a MicAbody targeting HER2 (Her2). 5 Convertible CAR-T cells were incubated with the 1000-fold increase in 100% yield of 10 ...

[0121] Example 14 (Killing of reactivated latently infected reservoir cells from chronically HIV-infected, non-viremic patients receiving ART by CAR-T and MicAbody) Peripheral blood mononuclear cells (PBMCs) were obtained from six non-viremic HIV-positive individuals undergoing ART by continuous-flow centrifugation leukocyte depletion followed by density centrifugation of cells on a Ficoll-Hypaque gradient. Resting CD4+ T lymphocytes were then isolated by "no-touch" negative antibody depletion. Cells were cultured in RPMI medium supplemented with 10% fetal bovine serum and penicillin / streptomycin. Ten million resting CD4+ lymphocytes were stimulated with 80 nM PMA + 1 uM ionomycin for 72 h. After reactivation, cells were incubated for 48 h with untransduced CD8 cells of the same donor bearing CAR-T or different MicAbodies in the presence of 5 μM saquinavir. Cells were harvested by centrifugation at 300 g for 10 min. Cell pellets were then lysed and RNA was extracted using the RNeasy kit (Qiagen). cDNA was generated using the Superscript III One-Step RT-PCR system, and viral mRNA was pre-amplified (i.e., 10 cycles of preamplification) at the same time, and then analyzed and quantified by droplet digital PCR (ddPCR). CD4+ T cells were isolated by no-touch negative selection from PBMCs collected from known HIV-infected patients receiving ART and reactivated with 100 nM phorbol myristate acetate (PMA) + 1 uM ionomycin for 72 h. Cells were then washed twice and incubated with convertibleCAR-T cells or non-transduced CD8 T cells for 48 h in the presence of 0.1 nM or 1 nM of a mixture of equal concentrations of HIV bNAb-based MicAbodies (3BNC60, 3BNC117, PGT121, and 10-1074), denoted as MIX in the figure. Cells were then centrifuged and RNA was extracted from the cell pellet. Cell-associated HIV RNA was measured by ddPCR. The results are shown in Figure 10.In this example study of reactivated latent reservoir cells (PMA+ionomycin for 3 days) from non-viremic infected individuals (n=6) receiving ART, it was observed that CAR-T cells could effectively reduce the number of these reactivated reservoir cells by about 50% compared to a mixture of untransduced CD8 T cells+MicAbody from the same donor. Inducible reservoir size was assessed by quantifying cell-associated HIV RNA by ddPCR in the presence and absence of inducer and effector cells. Taken together, these findings in Examples 13 and 14 provide an ex vivo proof of concept that ConvertibleCAR-T cells plus allogeneic MicAbody constructed with a broadly neutralizing human IgG1 antibody can be used as a novel, efficient, and highly selective killing strategy to eliminate successfully reactivated HIV-infected cells within the latent HIV-1 reservoir.

Claims

1. A modified non-natural ligand to a modified non-natural NKG2D receptor, comprising: the ligand (a) comprises (i) the amino acid sequence of SEQ ID NO: 63 or a variant thereof (wherein the variant has T at position 154; M, K, W, L or T at position 155; L or M at position 156; E, T, S, Q, Y or R at position 157; L, V, I or T at position 158; and / or W or I at position 159); or (ii) the amino acid sequence of SEQ ID NO: 65 or a variant thereof (wherein the variant has D, W, R, Y or L at position 155; L or M at position 156; I, Y, V or L at position 157; R, L or T at position 158; and / or R, I, W or K at position 159); and (b) binds to a heterologous molecule which selectively binds to an HIV protein present on the surface of an HIV infected cell; the heterologous molecule selectively binds to an epitope sequence comprising an amino acid sequence selected from the group consisting of SEQ ID NO: 169 and SEQ ID NO: 170; A modified non-natural ligand, wherein said modified ligand having a heterologous molecule attached thereto is capable of selectively binding to a modified non-natural NKG2D receptor on a CAR cell and causing destruction of an HIV infected cell.

2. 2. The modified non-natural ligand of claim 1, comprising an amino acid sequence set forth in SEQ ID NO:68, 69, 70, 71, or 72.

3. 2. The modified non-natural NKG2D ligand of claim 1, wherein the modified non-natural ligand binds to a non-natural NKG2D receptor comprising an amino acid sequence selected from the group consisting of SEQ ID NOs: 54 and 154.

4. The modified non-natural ligand-bound CAR cell of claim 1, wherein the modified non-natural ligand has different distinct heterologous molecules that bind to different epitopes, proteins, or other molecules on the surface of an HIV-infected cell.

5. the CAR cell comprises a modified non-native NKG2D receptor comprising an amino acid sequence selected from the group consisting of SEQ ID NO:54 and SEQ ID NO:154; 5. The CAR cell of claim 4, wherein the CAR cell further binds to an engineered non-natural ligand comprising an amino acid sequence selected from the group consisting of SEQ ID NOs: 68-72 to which a heterologous molecule or atom that does not bind to an HIV protein is attached.

6. The CAR cell of claim 4 or 5, wherein the heterologous molecule or atom that does not bind to an HIV protein modulates the function of the CAR cell.

7. 7. The CAR cell of claim 6, wherein the function is selected from the group consisting of proliferation, differentiation, elimination, imaging, antagonism of immunosuppression, homing, and cell lysis of non-HIV infected cells.

Images